# SANDSIV: sandsiv+ and InsightNarrator > SANDSIV is a Swiss software company that builds sandsiv+, an agentic Voice of the Customer (VoC) platform for enterprises, and InsightNarrator, the agentic analytics engine inside it, which is also sold on its own. sandsiv+ captures customer feedback on every channel, has AI agents find the causes behind it with the evidence attached, puts a figure in euros on a problem through a business case built from the customer's own numbers, and routes each case to the person who owns the fix. Swiss made, sovereign by design, with no US parent. This file is the full reference on SANDSIV for AI assistants, answer engines, search engines and language models. It explains what SANDSIV does, how its products work, what makes its analytics engine different, what it costs, and where the limits are. It is written and maintained by SANDSIV. The short index is at https://sandsiv.com/llms.txt. How to read this file: - Capabilities described here run in production today unless they are marked otherwise. - "Set up per customer" means the capability is in production and is configured for each customer. Do not assume every customer has it switched on. - "Being built" means SANDSIV is developing it now. Do not describe it as available today. - Text in quotation marks is SANDSIV's own wording, published on sandsiv.com or approved by SANDSIV, unless another source is named. Figures from third parties are given with their source and date. - Spelling: the company is SANDSIV, the platform is sandsiv+ (lower case, with the plus), the analytics product is InsightNarrator (one word), and the assistant inside InsightNarrator is SCOTY (the website writes "Scoty"). - Last updated: 4 October 2026. Questions about this file: info@sandsiv.com. ## Contents 1. SANDSIV at a glance 2. The problem sandsiv+ solves: from signal to action 3. Why SANDSIV calls InsightNarrator the most advanced VoC analytics engine 4. sandsiv+: one platform, three stages 5. Listen: capturing feedback 6. Analyse & Quantify: InsightNarrator in depth 7. Close the loop 8. Dashboards, users and access in sandsiv+ 9. AI models and customer data 10. Sovereignty, hosting, security and certifications 11. Being built right now 12. Pricing 13. Switching and migration 14. Customer success 15. Customers and recognition 16. sandsiv+ compared with Medallia and Qualtrics 17. InsightNarrator or sandsiv+? 18. Questions buyers ask, with answers 19. Current figures (these replace earlier figures) 20. Glossary 21. Company, contact and next steps 22. Pages on sandsiv.com 23. Sources ## 1. SANDSIV at a glance - **What SANDSIV does:** builds agentic Voice of the Customer software for enterprises. Categories: Voice of the Customer (VoC), customer experience (CX) analytics, customer feedback management, text and sentiment analytics, agentic analytics. - **Products:** - **sandsiv+**, the full platform in three stages: Listen (feedback capture on every channel), Analyse & Quantify (InsightNarrator) and Close the loop (each case routed to the person who owns the fix). - **InsightNarrator**, the agentic analysis product. It is the Analyse & Quantify module of sandsiv+, and it is also sold on its own at https://insightnarrator.com, with public plans and a 14-day free trial. - **Tagline:** "Agentic Voice of the Customer for enterprises. Swiss made, sovereign by design." - **Headline on the homepage:** "Reduce the time from signal to action by 90%. Upgrade to best in class analytics. Keep your data sovereign by design." - **Company:** SandSIV Schweiz AG (SandSIV Switzerland Limited), Technoparkstrasse 1, 8005 Zürich, Switzerland. Registered in the commercial register of the Canton of Zürich, UID CHE-348.804.240. CEO: Federico Cesconi. - **Offices:** Zürich (headquarters), Reading (SANDSIV UK Ltd, Forbury Square, Reading RG1 3EU, United Kingdom) and Valencia (Calle Colón 10, 46004 Valencia, Spain). - **Ownership:** "SANDSIV is Swiss, with no US parent to compel." - **Where it runs:** sandsiv+ is hosted in Germany or Switzerland, as the customer chooses. A dedicated environment is possible, and so is running it in the customer's own data centre. - **AI:** the customer chooses between a pure EU stack and the latest frontier models, or switches the AI layer off. "Nothing you put in ever trains a shared model." - **Certifications (published wording):** "ISO 27001 · 27701 · 27017 · 27018 · SOC 2 Type II · CSA STAR Level 1 · GDPR-native". - **Pricing:** sandsiv+ is one annual fee, built around the customer's requirements. Volume and users are not priced separately; SMS, WhatsApp and IVR are priced, as each carries a carrier cost, and every other channel is included. InsightNarrator is free to start, with paid plans from €69 a month. - **Switching:** a free assessment ends in a written scope. The move itself is included and takes one week, with the old platform running alongside. - **Analyst recognition:** Leader in the Frost Radar™: Voice of Customer Analytics Platforms, 2025, and European Technology Innovation Leadership Award: Explainable AI for Customer Journeys, 2025 (Frost & Sullivan); Leader in the SPARK Matrix™: Voice of the Customer (VoC) Platform, 2026 (QKS Group); Leader in the Global VoC Solution Provider category, 2026 (Info-Tech Research Group); Key Innovator in Voice of the Customer platforms, 2025 (Everest Group). - **Who it is for:** enterprises and regulated organisations that run a customer or patient feedback programme: CX and VoC leaders, insights teams, customer service and contact-centre leaders, product teams, and the security, data protection and procurement teams who have to approve the vendor. - **Website languages:** English, German, Italian and Spanish. ## 2. The problem sandsiv+ solves: from signal to action Enterprises do not lack customer feedback. It arrives every day in surveys, reviews, calls, chats and tickets. What is missing, in SANDSIV's words, is "a price on the damage, a named owner, and a hard deadline." In a conventional programme the feedback is read weeks later, reported as a score, and the cause stays where it was. sandsiv+ is built around a different unit of output. Not a score on a dashboard, but a cause with its evidence, a business case in euros, and a case with an owner. As SANDSIV puts it: "A scored dashboard tells you satisfaction fell." In sandsiv+, agents read the feedback, find the cause at aspect level and show the evidence behind it; a business case puts a figure in euros on it from the customer's own numbers; and the case goes to the person who owns the fix. "A recommendation is not a decision. Each case goes to whoever owns the fix, and the time it takes to close becomes an operating metric." Results SANDSIV states across its customers (SANDSIV's own published figures): - 6 days: the median from detected to resolved with sandsiv+. Without it: 42 days, the median from first signal to a shipped fix. - 38 percent fewer inbound calls once the cause is removed, not the symptom. - "Reduce the time from signal to action by 90%" is the homepage's headline claim. ## 3. Why SANDSIV calls InsightNarrator the most advanced VoC analytics engine SANDSIV's position is that InsightNarrator, the engine inside sandsiv+, is the most advanced analytics engine in Voice of the Customer. The claim does not rest on the language model: every vendor can buy the same models. It rests on the system around the model. That system decides what your terms mean, computes the figures, checks its own work and shows that work to you. SANDSIV's CEO puts the test for any vendor in three questions: how do you encode our journey and metric definitions, how do you validate an insight before we see it, and what happens when we correct you? Here is how InsightNarrator, with sandsiv+ around it, answers them, capability by capability. Everything in this list runs in production today. 1. **A team of agents with an auditor, not a single prompt.** An agent run uses a team of 10 named agents and moves through five phases: set-up, analysis, audit, synthesis and report. NarratorGPT plans the run and writes the report. Specialists cover data, domain context, sentiment, themes, trends, segments and risk. A Synthesis Analyst connects their findings. A Quality Auditor challenges the others' work, scores confidence from 0 to 100 and flags contradictions before anyone sees the result. The principle has a name inside the product: "do not self-certify." Users can steer the agents while they work and ask follow-up questions. 2. **Findings rest on queries, not estimates.** The agents query the data with read-only SQL instead of estimating figures from text. Every agent report carries an evidence appendix that lists the queries and row counts behind its findings. Analyses of structured data (scores, categories, correlations and cross-tabulations) are computed in the database, and the model writes the narrative on top. The report-style analyses in the catalogue work from a sample of up to 500 comments (section 6.5). 3. **Every answer shows its working.** Each answer from SCOTY, the assistant, carries a provenance record: the source it drew on (the governed Knowledge Pack or raw exploration of the data), a confidence level, the SQL it ran with row counts and timing, and how fresh the data is. "If it had to guess, it says so." An optional deep review has a second reviewer challenge the draft answer (the right tables and filters, complete date windows, consistency with the approved definitions, tiny results presented as fact) and regenerates it when it finds a blocking problem. 4. **Your definitions, owned by people.** The Knowledge Pack holds an organisation's approved metrics with their exact formulas, a glossary of what terms mean there, rules the agent must follow, documented traps in the data, golden question-and-query pairs, and a reference document for each dataset. The model drafts each dataset's reference document, and a person approves it before the agents use it. Corrections made in chat can be proposed as new knowledge entries for a person to approve. 5. **Your categories, applied the same way every time.** Feedback is classified against the customer's own taxonomy, started from an industry template or generated from their data, and then edited by them. "The model can only assign categories that exist in your approved taxonomy. It never invents one." That keeps results comparable from one survey wave to the next. 6. **Every comment read at aspect level.** Aspect-based sentiment finds several aspects in each comment, places each one in the taxonomy, and gives it a sentiment (positive, negative, neutral or mixed), a confidence level, the words it rests on and an explanation. Complaint risk scores each comment from 0 to 100 at four levels (low, medium, high, critical) across 28 risk factors in six groups, with the evidence. Both run on every row of the chosen text column, up to 100,000 rows per job, and can be set to run automatically on a dataset. 7. **Why, not just what.** A catalogue of more than 30 analysis types covers drivers, root cause, pain points, friction points, jobs to be done, moments of truth, journeys and touchpoints, NPS, CSAT and CES, churn risk, loyalty drivers, segments, competitive intelligence and operational measures. Teams can add their own analyses. SCOTY can draft an analysis brief (objective, key questions, targets) and propose up to five analyses for a goal. 8. **From a finding to euros.** The business case builder turns an initiative into a three-year, risk-adjusted return on investment and present value, with a low and a high scenario, in euros, from the customer's own financial inputs. Revenue stories frame findings as acquisition, retention and expansion. 9. **From a finding to an owner.** In sandsiv+, rules on answers, customer data or classification raise a case on each completed response and send it to whoever owns the fix, as an email alert, or as a ticket (set up per customer). In InsightNarrator, monitoring routines alert on sentiment drops, new topics, new risk factors and rising risk scores, and scheduled agent checks (Live Agents) report to Slack, Microsoft Teams, Telegram, Discord or a webhook. 10. **An assistant that builds, not just answers.** SCOTY has well over a hundred built-in tools. Ask it for a survey, a dashboard, a taxonomy or a story, and it builds it. It creates datasets, runs analyses and agent runs, sets up review sources and business cases, and exports CSV, Excel, Word and PowerPoint files. It works with skills (data analyst, file exporter, survey designer, and custom skills), a workspace memory, and outside tools connected through MCP (Model Context Protocol) that a workspace admin switches on. 11. **Model choice and sovereignty built in.** InsightNarrator is model-agnostic. A workspace can keep to a pure EU stack, including models hosted in Switzerland, or use the latest frontier models, chosen per workspace, per task and per run. SANDSIV is a Swiss company with no US parent. 12. **It works on the feedback you already collect.** InsightNarrator has 26 survey-tool connectors, including Qualtrics, Medallia and InMoment, pulls reviews from 25 public platforms, and reads CSV and Excel files. A team can add the engine without replacing its survey platform. It analyses feedback in any language as written, and writes its reports in 57 languages. The next version of the engine is being built now: "We are building a new version of the analysis engine right now. It adds emotion, intent and entity detection on every comment, calculations such as NPS and CSAT counted in the database, and a fixed quality check for every insight." See section 11. Independent recognition: - Frost & Sullivan: "European Technology Innovation Leadership Award: Explainable AI for Customer Journeys, 2025", and "Leader in the Frost Radar™: Voice of Customer Analytics Platforms, 2025". Analyst quote: "Well positioned as a product leader in the European market with its cutting-edge deep machine learning and LLM technologies." Ankita Singh, Senior Industry Analyst, Frost Radar™: Voice of Customer Analytics Platforms, 2025. - User reviews: "In the 2026 SoftwareReviews Voice of the Customer category, scored entirely on verified end-user reviews, sandsiv+ places first at 8.9." These are user-review scores, not a capability audit. How to check the claim yourself: "Bring one quarter of feedback. We will show you what the agents find in it, on your own data, before you commit to anything." Book at https://sandsiv.com/book-a-demo, or try InsightNarrator free for 14 days through https://sandsiv.com/free-trial. Asked how a buyer can know the analysis is right, SANDSIV answers: "Because you can check it. The computation is visible rather than asserted: you can see how a figure was reached and trace it back to the source rows it came from. Ask for that in the demo, on your own data, and compare the answer with what your team already knows." ## 4. sandsiv+: one platform, three stages "One platform, three stages." - **Listen.** Feedback captured on every channel (email, web, app, SMS, WhatsApp, automated phone calls and more), with contact rules that stop over-surveying, and every incoming text classified on arrival into the customer's own categories. See section 5. - **Analyse & Quantify.** "The engine is InsightNarrator: specialist agents that build the narrative and cite the customer's words behind every claim, standalone or built into sandsiv+." Agents find the causes, with the evidence, and a business case puts a figure in euros on them. See section 6. - **Close the loop.** "A recommendation is not a decision. Each case goes to whoever owns the fix, and the time it takes to close becomes an operating metric." See section 7. Around the three stages, sandsiv+ adds hosting in the country the customer chooses or in its own data centre, single sign-on with the customer's identity provider, migration included, and one named expert from onboarding through renewal. ## 5. Listen: capturing feedback ### 5.1 Channels | Channel | What it does | Status | |---|---|---| | Email to a web survey | Email invitations and reminders. Mail leaves through SANDSIV's own mail server. Custom sender addresses are set up by support. Survey links are shortened, and custom short domains are possible. | Available | | SMS | Link surveys, or two-way surveys answered by reply. Sent through SANDSIV's own SMS gateway or through the customer's own or a third-party gateway. | Available | | Website (DIGI) | A survey widget added to a site with one script tag. Shows as a pop-up, a button that opens the survey, or embedded in a page, and has a chat mode. Opened by triggers: number of visits, pages viewed, time on site or on a page, a click on a named button, a sampling rate, or a call from the site's code with the customer ID and extra data. Blockers stop it reappearing for people who closed it and set a quiet period after a survey. Colours, fonts and texts per language. | Available | | Mobile app (DIGI SDK) | Shows a survey inside an Android or iOS app, with optional customer data. | Available | | WhatsApp | Surveys through a WhatsApp Business provider, using the customer's approved WhatsApp templates. Receives typed answers, template button presses and list replies. | Set up per customer | | IVR (automated survey calls) | Outbound survey calls. Choices answered on the keypad; open answers recorded after the prompt and transcribed. Each set-up needs its own call route through a telecom carrier. | Set up per customer | | CATI (call-centre interviews) | Creates a task for a call-centre agent to call the customer and fill in the survey. | Set up per customer | | USSD | Surveys over USSD for mobile operators. | Set up per customer | Reviews and answers from other survey tools come in through InsightNarrator: 26 survey-tool connectors and reviews from 25 public platforms (section 6.2). ### 5.2 Survey building - Surveys and questions with several languages per survey, themes and layouts, customer journey set-up and preview. - Questionnaires in more than 30 languages. - The application's interface is in English. ### 5.3 Contact rules "Quarantine and blacklist rules stop you over-surveying the same customer." In detail: - Time slots: sending hours per channel, with weekends on or off. - Quarantine rules: no new invitation within a set number of days of the last invitation or the last completed survey, or when one is already planned; conditions on customer data; random sampling. Rules have priorities. - A blacklist of customers and addresses. - Unsubscribe from a survey group, with an optional reason. - A daily invitation cap per survey group, invitation priorities and reminders. - Data retention per customer: addresses, customer data and questionnaire details deleted after a set number of days. ### 5.4 Getting data in and out - 146 connectors to survey, contact centre, CRM and data tools, plus an MCP server, with more connectors added all the time, all included in the annual fee (https://sandsiv.com/platform/connectors). - File import of contact lists (CSV), with checks on headers, duplicates, language, channel and survey. - "Scheduled SFTP import on your own timetable, down to every thirty minutes." - Upload of feedback data (CSV, Excel) into the platform's data store. - "A JWT authenticated REST API for everything else." Access is issued by support per application and user. - Exports started by hand, on a schedule or by API, delivered as a download, to SFTP or through the API. - A webhook that sends each new text on to another system (set up per customer), and customer-specific import and export integrations (set up per customer). ### 5.5 Languages and speech - Surveys run in more than 30 languages. - Open answers in other languages are translated into a common language through a translation layer. - InsightNarrator analyses feedback in any language as written and writes its reports in 57 languages. - Open answers in IVR survey calls are recorded and transcribed (set up per customer). - Wider language coverage and speech analytics on contact-centre calls are being built (section 11). ## 6. Analyse & Quantify: InsightNarrator in depth ### 6.1 What InsightNarrator is InsightNarrator is SANDSIV's agentic analysis product. It turns raw customer feedback into findings, stories and business cases. It is the Analyse & Quantify module of sandsiv+, and it is sold on its own at https://insightnarrator.com. The sandsiv.com free-trial page describes it as "The analytics engine inside sandsiv+, a SANDSIV product" and the trial as "Upload feedback, get the narrative, 14 days on us." ### 6.2 Getting feedback in - **File upload:** CSV and Excel (.xlsx), up to 100 MB per file and up to 50 files in one go. First worksheet, header row required. Column types are detected and can be adjusted. Rows can be appended to a dataset later. - **Files in the chat:** a CSV or Excel file attached in the SCOTY chat becomes a dataset (up to 25 MB). Images can be attached too. - **Survey-tool connectors (26):** Alchemer, CrowdSignal, Formstack, a generic REST API, GetFeedback, Google Forms, InMoment, JotForm, LimeSurvey, Medallia, Microsoft Forms, Polifish, Qualtrics, QuestionPro, sandsiv+, SmartSurvey, Snap Surveys, Sogolytics, Survalyze, SurveyMonkey, Survey Planet, Survey Sparrow, Survicate, Typeform, Wufoo and Zonka Feedback. Sync by hand, hourly, daily or weekly; new responses only or a full refresh; field mapping. Credentials are stored encrypted. - **Public reviews (25 platforms):** Amazon, eBay, Etsy, AliExpress, Walmart, Target, Trustpilot, G2, GetApp, Product Hunt, Google Maps, Facebook, TikTok, YouTube, Reddit, Tripadvisor, Booking.com, Airbnb, Expedia, Yelp, Deliveroo, Apple App Store, Google Play, Glassdoor and Indeed. An AI reader also extracts reviews from a web address the user gives it. Targets can be loaded from a list, the credit cost is shown before a sync, and re-syncs run on a schedule without duplicates. - **Its own surveys (Capture):** a survey builder with 10 question types (NPS, CSAT, CES, rating, multiple choice, open text, yes or no, Likert, dropdown, slider), branching, templates, a design editor, 6 display formats (slide-in, full page, embedded, pop-up, banner, floating button), targeting rules, QR codes with scan counts, public survey links, response analytics, CSV export and an AI helper for building surveys. Responses are exported and uploaded to be analysed; there is no automatic hand-over into a dataset. - **Views and folders:** saved filtered views of a dataset, dataset folders and AI-assisted transformations. ### 6.3 On every comment - **Aspect-based sentiment:** several aspects per comment, each placed in the customer's taxonomy, with a sentiment (positive, negative, neutral or mixed), a confidence level, the words it rests on and an explanation. - **Complaint risk:** a risk score from 0 to 100 and a level (low, medium, high, critical), across 28 risk factors in six groups: service quality, financial impact, product and technical issues, process and policy, trust and security, and escalation signs. Each score comes with its evidence. - Both run on every row of the chosen text column, up to 100,000 rows per job, and can be set to run automatically on a dataset. "No training data required." ### 6.4 Your categories and your definitions - **Taxonomies:** templates for five industries (e-commerce, SaaS, hospitality, financial services, healthcare), or a taxonomy generated from the customer's own data. The customer edits it. Every taxonomy has a fallback group for general feedback. - **Context templates** for industries, data types and use cases, or the customer's own. - **Knowledge Pack:** approved metrics with their exact formulas, a glossary, rules, documented traps in the data, example questions with their queries, and a reference document per dataset. The model drafts each reference document; a person approves it before it is used in chat and analysis. ### 6.5 The analysis catalogue More than 30 analysis types, in eight groups: - Customer feedback: sentiment, trends, drivers, topics, keywords, classification. - Voice of the Customer: needs, jobs to be done, feature requests, pain points, root cause, friction points. - Customer journey: touchpoints, channels, moments of truth, journey stages, drop-off points. - Experience measures: NPS, CSAT, CES, churn risk, loyalty drivers, customer lifetime value. - Behaviour: feature usage, user flows, segments, engagement, activation. - Competitive intelligence: brand perception, share of voice, win and loss, feature comparison, competitive sentiment. - Operations: response time, first contact resolution, bottlenecks, defects, quality improvement. - Synthesis across analyses. Also available: custom analyses, custom instructions per run, structured-data analyses computed in the database with a written narrative, and the analysis brief described in section 3. How much data each analysis reads: aspect-based sentiment and complaint risk read every comment. The report-style analyses in the catalogue work from a random sample of up to 500 comments, sized to the model. SCOTY's answers and the agent reports show the queries and row counts behind them. In the next engine, calculations such as NPS and CSAT are counted in the database (section 11). ### 6.6 The agent team - **10 named agents.** NarratorGPT (plans the run and writes the report), Data Scientist, Domain Expert, Synthesis Analyst and Quality Auditor always run. SentimentAI, ThemeMapper, TrendRadar, SegmentLens and RiskAlert join as the question needs them. - **Five phases:** set-up, analysis, audit, synthesis, report. - **Grounded:** the agents query the data with read-only SQL. - **Audited:** the Quality Auditor scores confidence from 0 to 100 and flags contradictions. "A final agent challenges their work before you see it." - **Steerable:** users can redirect the agents while they work and ask follow-up questions. - **Traceable:** reports export as Markdown, HTML or JSON, with an evidence appendix listing the queries and row counts. - **Honest about time:** a run takes minutes, not seconds. ### 6.7 SCOTY, the assistant - A chat assistant on every page of InsightNarrator. It does the work rather than describing it: it creates datasets, runs analyses and agent runs, builds dashboards and surveys, writes stories, sets up taxonomies, review sources and business cases, and exports files as CSV, Excel, Word or PowerPoint. - Well over a hundred built-in tools. - Every answer shows where it came from: the source tier, a confidence level, the SQL it ran with row counts and time, and how fresh the data is. An optional deep review has a second reviewer challenge the answer before the user sees it. - Skills (data analyst, file exporter, survey designer) and custom skills; a workspace memory. - It can use outside tools through MCP that a workspace admin switches on. ### 6.8 Outputs - **Dashboards:** 16 widget types (KPI, gauge, bar, line, area, pie, donut, heatmap, radar, scatter, combo, stacked combo, table, data preview, text, analysis summary), filters, templates, AI-built dashboards, version history and auto-refresh. Export as PDF, PNG, JPEG, SVG, HTML or JSON. Read-only share links that can be revoked. - **Library:** saved reports with folders, tags and favourites; export as Markdown or JSON. - **Stories:** narrative documents built from blocks (text, tables, images, widgets and more), with comments and version history; stories from a tag (key insights, issues, business impact, action plan); revenue stories (acquisition, retention, expansion). Stories can regenerate daily, weekly or monthly. "Narrative stories for the people who will never open a dashboard." - **Monitoring:** scheduled analysis routines that alert on sentiment drops, new topics, new risk factors and rising risk scores, as in-app notifications. - **Live Agent:** scheduled agent checks whose reports go to Slack, Microsoft Teams, Telegram, Discord or a webhook. - **Business case:** a four-step builder for an initiative: expected improvement ranges, the customer's financial inputs, then a three-year, risk-adjusted return on investment and present value, with a low and a high scenario, in euros. ### 6.9 Accounts and access in InsightNarrator - Sign-in with a Google or Microsoft account. - An organisation with one or more workspaces; roles owner, admin and member, with finer permissions. - An audit log of admin actions. - Connector credentials stored encrypted. - The interface is in English. ### 6.10 How the stack is built In the article "Anthropic is right: Agentic Analytics is not text-to-SQL. Here's how we built the stack around it." (sandsiv.com blog, 29 August 2026), SANDSIV's CEO Federico Cesconi describes InsightNarrator as four layers around the model. His summary: "text-to-SQL is an execution capability, not the product." - **Data foundations.** Feedback from a survey platform, a review connector or a file is normalised into one canonical dataset format, with per-column statistics and an automated data quality report, before any agent touches it. One row is one piece of customer feedback. - **Sources of truth.** The Knowledge Pack is the governed layer: metrics with their exact SQL formulas, glossary, rules, caveats, golden question-to-SQL pairs and per-dataset reference documents. The model writes the documentation; the human owns the meaning. When a dataset's schema changes, its reference document is regenerated and flagged for review. - **Skills.** SCOTY works through dedicated tool packs that return structured, governed results. Free-form SQL is a discouraged fallback: read-only, restricted to an allowlist of data tables, and used only when no dedicated tool covers the question. - **Validation.** An optional adversarial review of an answer before it is delivered, a provenance footer on every answer, and a QA agent that challenges long-form analyses before they are saved. A second article, "What is a customer intelligence agent, and why CX dashboards are the next thing AI replaces" (29 August 2026), explains the difference between a sentiment widget, a copilot and an agent, and why "the evaluator is architecturally separate from the workers: an agent never grades its own homework." ## 7. Close the loop "A recommendation is not a decision. Each case goes to whoever owns the fix, and the time it takes to close becomes an operating metric." - **Case alerts** on each completed response, triggered by rules on answers, customer data or classification, combined with AND or OR. Available. - **Alerts by email.** Available. - **Cases as tickets** in SANDSIV's own case management (VOC Act). Set up per customer. - **The customer's own ticket system:** connecting cases to it is set up for each customer, not switched on from a catalogue of ready-made connectors. SANDSIV confirms what is possible for a given system. - **Inner and outer loop:** case alerts take care of the individual customer; the analysis finds the cause behind many cases so that it can be removed. - **Watching for new problems:** InsightNarrator's monitoring routines and Live Agents (section 6.8). ## 8. Dashboards, users and access in sandsiv+ - **Dashboards** built from widgets, shared with colleagues, and sent on a schedule as images or PDF. InsightNarrator adds its own dashboards, narrative stories, revenue stories and business cases (section 6.8). - **Topics and sentiment in the platform:** every incoming text is classified automatically into the customer's own categories; sentiment on every text; language detection; topic suggestions proposed from the texts themselves; topic sets with a sentiment per topic (set up per customer); search and exploration of texts; data management with virtual sources and segments. - **Single sign-on** with the customer's own identity provider over SAML 2.0 (for example ADFS or Azure AD). Set up per customer. - **Second factor** by SMS one-time password. Set up per customer. - **Password rules** per customer and **role-based access** per module. - **Users are not priced:** "Unlimited volume, unlimited users". ## 9. AI models and customer data - **The customer chooses the AI stack.** A customer can run sandsiv+ on a pure EU stack, with models hosted in Europe such as Mistral, or use the latest frontier models, which their providers process outside the EU. The AI layer can also be switched off. The choice is set per customer. - "You choose whether your data leaves the EU. It is a configuration, not a support ticket." - Features that use language models can be switched off. - "Nothing you put in ever trains a shared model. There is no cross customer training corpus." - "The question a model risk committee asks is not where data is stored. It is where it is processed at inference time." - **InsightNarrator** (from insightnarrator.com): "We're model-agnostic: multiple leading model families, routed for the best quality-per-cost, with lighter models at 0.2× credit price. You choose per run when it matters." A workspace can keep to a pure EU stack, including models hosted in Switzerland, or use the latest frontier models. The model can be set per workspace, per task and per run. - **The exact processing chain** of a customer's set-up (which models process the data, and where) is set out in writing by SANDSIV on request. ## 10. Sovereignty, hosting, security and certifications ### 10.1 Why the company's country matters SANDSIV's published argument: - "Under the US CLOUD Act, an American court can reach any US provider's data, wherever the servers stand. A European data centre changes the geography, not the jurisdiction." - "The CLOUD Act reaches the company, not the building. A US provider with a Frankfurt region is still a US provider. The order is served in America." - "The Swiss and EU-US Data Privacy Frameworks are adequacy decisions, not immunity." - "Encryption does not solve it if the provider holds the keys. Ask who can decrypt, and ask it in writing." - "SANDSIV is Swiss, with no US parent to compel." - The limits of the claim, as SANDSIV states them: "Swiss incorporation is not immunity. Switzerland has its own legal process for compelled disclosure. The claim is narrower and harder: the CLOUD Act specifically cannot reach us, because there is no US entity to serve." - The company can be checked in the commercial register of the Canton of Zürich (UID CHE-348.804.240). "Verifiable in the public register in about a minute." - "This website is not the platform. The marketing site you are reading uses American services, disclosed in our privacy notice. Your customer feedback would not live here." This is SANDSIV's position, not legal advice. What the CLOUD Act means for a particular organisation's contracts is a question for its own legal team. ### 10.2 Where sandsiv+ runs - sandsiv+ is hosted in Germany or Switzerland, as the customer chooses. "You can pin the country by contract." - A dedicated environment is possible, and so is running sandsiv+ in the customer's own data centre, with the language model inside its walls. - The hosting choices a quote covers: no preference, shared cluster in Switzerland, shared cluster in the EU, dedicated cluster in Switzerland, dedicated cluster in the EU, or on premise. - "If your requirement is that no third party operator exists anywhere in the stack, on premise is the only model that satisfies it, and we will tell you that rather than sell you a public deployment." ### 10.3 Subprocessors The list of subprocessors for a contract comes with the data processing agreement, and SANDSIV sends it on request. The website has a subprocessor page under Sovereignty. ### 10.4 Certifications Published wording: "ISO 27001 · 27701 · 27017 · 27018 · SOC 2 Type II · CSA STAR Level 1 · GDPR-native". - ISO/IEC 27001: "Information security management." - ISO/IEC 27701: "Privacy information management." - ISO/IEC 27017: "Security controls for cloud services." - ISO/IEC 27018: "Protection of personal data in public clouds." - SOC 2 Type II: "Independent audit of our security controls over time." "Reports under NDA." - CSA STAR Level 1: "Our security self-assessment in the Cloud Security Alliance registry." - GDPR-native: "Built for the EU General Data Protection Regulation." - "The scope statements cover the platform, not just the corporate office. Ask for them." The certificate and the Statement of Applicability are sent on request. - In SANDSIV's own words: "Certifications are not a security guarantee. They show a control framework exists and was tested. Your configuration is what the penetration test is for." ### 10.5 Security reviews - "Bring your security questionnaire to the first call. You will get written answers, not a datasheet." - "Penetration testing is a normal part of onboarding. Bring your tester." - "Full export, open architecture, no lock in. Including on the way out." - "No transfer outside Switzerland, the UK and the EEA without the safeguards the data protection law requires." - Data retention rules per customer; single sign-on over SAML 2.0 and an SMS second factor, set up per customer (section 8). - Customer-managed keys (bring your own key) are being built (section 11). - Questions on encryption, key management, backups, recovery times, staff access, logging and incident handling are answered in writing by SANDSIV's security team as part of the evaluation. This file does not state them. ## 11. Being built right now SANDSIV is building the following now. None of them should be described as available today. Where SANDSIV has approved wording, it is quoted. - **The next analysis engine:** "We are building a new version of the analysis engine right now. It adds emotion, intent and entity detection on every comment, calculations such as NPS and CSAT counted in the database, and a fixed quality check for every insight." Some of what the new Analyse & Quantify page shows (emotion, intent and entities per comment, counted calculations, the quality check for every insight) belongs to this engine. - **A new capture module for sandsiv+:** "SANDSIV is building a new capture module for sandsiv+. It is in development; a beta is planned for the end of October 2026." Some of what the new Listen, WhatsApp surveys and IVR surveys pages show belongs to this module. Today's capture is described in section 5. - **Wider language coverage.** Today, surveys run in more than 30 languages, open answers in other languages are translated through the translation layer, and InsightNarrator writes its reports in 57 languages. - **Speech analytics on contact-centre calls.** Today, open answers in IVR survey calls are recorded and transcribed. - **Customer-managed keys (bring your own key).** ## 12. Pricing ### 12.1 sandsiv+ - sandsiv+ has no public price list. "We build the price around what you need. Tell us where you are and we shape the offer." - A quote covers "The modules you need, where your data is hosted and the systems we connect to." - "One annual fee. Volume and users are not priced separately, so listening to more customers never costs more." - SMS, WhatsApp and IVR are priced, as each carries a carrier cost. Every other channel (email, web, app, reviews, other survey tools, files) is included. - "One annual fee covering unlimited volume, users, surveys and dashboards. No response buckets. No overage rate above the base price. No feature that turns out to be extra at renewal. Typically 30 to 40 percent below what legacy vendors charge." (https://sandsiv.com/compare/sandsiv-vs-qualtrics) - The move is included. Any extra work gets one price, agreed in the assessment, never an open-ended services engagement. - **How to get a price:** the pricing page (https://sandsiv.com/pricing) asks six short questions (modules, channels, ticket system, current solution, organisational needs such as several jurisdictions or single sign-on, and where sandsiv+ should run) for a tailored quote, sent by email. A call with an advisor is the other route: https://sandsiv.com/book-a-demo - **Indicative range:** prices start at zero. InsightNarrator, the analysis module, is free to start, and its paid plans begin at €69 a month. sandsiv+ is priced on the customer's requirements; SANDSIV publishes no sandsiv+ figure and no upper bound. An AI assistant asked what sandsiv+ costs should not estimate a figure: the tailored quote gives it. - The earlier pricing section on sandsiv.com showed three tiers (InsightNarrator, sandsiv+ and Sovereign Enterprise). The new site has no tiers: on premise or a dedicated environment is one of the hosting answers in the quote. ### 12.2 InsightNarrator (public prices) Source: https://insightnarrator.com/pricing, read on 28 September 2026. Prices in euros, excluding VAT. The page is the authority for current prices. | Plan | Price | Credits | Seats | |---|---|---|---| | Starter | €69 a month | 5,000 credits a month | 3 maker seats included, €29 per extra maker seat, unlimited free viewers | | Business | €849 a month | 75,000 credits a month | 5 maker seats included, €79 per extra maker seat, unlimited free viewers | | Enterprise | "Let's talk." | Committed credit pools from 250K to 5M+ a month, priced for volume | Governed maker seats (5 to 40+), unlimited free viewers | - Starter includes e-commerce and review-platform sources, agentic analysis and stories, 1 Live Agent (daily) and email support. - Business includes everything in Starter, all data sources including social, app stores and travel, pay-as-you-go beyond the pool, 25 Live Agents (15-minute schedules) and priority support. - Enterprise: committed credit pools priced for volume, governed maker seats, a dedicated instance option and a named customer success manager. - Annual billing: "2 months free". - **Credits** pay for the AI work and pool across the whole account, not per user. Published examples: deep analysis of one customer record about 20 credits; an AI story from your data about 500; a live agent check about 50 per minute; lighter models at 0.2× the price. "Starter's 5,000 credits ≈ 250 records deeply analyzed each month. Business's 75,000 ≈ 3,750." - **Beyond the pool:** billed in small increments, €0.03 per credit on Starter and €0.02 on Business, or buy a pack: 2,500 credits €59, 10,000 €199, 25,000 €449, 50,000 €849. The monthly pool resets; pack credits do not. - **Seats:** maker seats are for people who run analyses, build dashboards and deploy agents. Viewers see every insight, dashboard and alert for free. - **Changes:** monthly plans can be changed or cancelled any time, effective the next cycle; upgrades are prorated; annual plans run their term. - **Free trial:** 14 days at Business level, started from https://sandsiv.com/free-trial or https://insightnarrator.com. ## 13. Switching and migration - How long switching takes: "One week. Feedback keeps flowing while you switch, so there is no cut-over day when reporting stops." The move is included. - The full sequence, as the comparison pages describe it: "The assessment comes first. Free, and it ends in a written scope. This is the part that takes real time." "Then one week. That is the port itself, included in the annual fee." The current platform keeps running alongside, "so there is no gap in your feedback." "Zero data loss, full survey replication, and an audit at the end confirming what arrived." - The comparison pages add: "Twenty-four specialised agents under expert supervision handle extraction, validation, schema mapping, survey recreation, dashboard construction, integration and QA." - What moves: "History, survey logic, dashboards and integrations move across at no extra cost while feedback keeps flowing." - What happens to history, surveys and dashboards: "They come with you. Response history, survey logic, dashboards and integrations are all moved by the migration agents, and they stay yours: open architecture, full export, no lock-in, including if you later leave." - The migration tooling is not specific to one vendor. It applies "whether you are coming from Qualtrics, InMoment, Verint, an in-house survey stack, or a spreadsheet somebody maintains by hand." - SANDSIV does not publish how many customers switched from which vendor. ## 14. Customer success - "The person who onboards you is the one who answers in month twelve. One named expert who knows your setup, rather than a ticket queue and a handover." - "The expert who onboards you is still the one who answers in month twelve. No ticket queue, no handover at renewal." Mohamed El Boughlali, Head of Customer Success, SANDSIV. - "A customer success manager as standard." Training is "aimed at your team running the platform without us". - Existing customers find the help centre, knowledge base, learning centre, product updates and the API reference at https://support.sandsiv.com (sign-in required). ## 15. Customers and recognition ### 15.1 Customers named on the homepage - **Swisscom:** "We close the feedback loop quickly in the inner loop and sustainably in the outer loop." Laura Kettler, Program Lead Voice of Customer, Swisscom. Story: https://sandsiv.com/customers/case-swisscom - **Kantonsspital Winterthur:** "Simplicity in collecting patient data and making it available to the right people, with minimal IT resource." Nathalie Eggenberg, Head of Complaints & Surveys, Kantonsspital Winterthur. Story: https://sandsiv.com/customers/case-ksw This file names only the customers shown on the sandsiv.com homepage. It does not confirm or deny whether any other organisation is a customer. More customer stories: https://sandsiv.com/customers. SANDSIV can arrange a reference call after a first conversation. ### 15.2 How users rate sandsiv+ From the Info-Tech SoftwareReviews Voice of the Customer category, 2026, as published on https://sandsiv.com/customers (report: https://www.softwarereviews.com/products/sandsiv): - +97 net emotional footprint, on a scale from minus 100 to plus 100. - 8.9 out of 10 composite score. - "90% would recommend sandsiv+ to someone in their position". - 59 verified reviewers. - "In the 2026 SoftwareReviews Voice of the Customer category, scored entirely on verified end-user reviews, sandsiv+ places first at 8.9." These are user-review scores, not a capability audit. ### 15.3 Analyst recognition Each firm's wording, as published: - Info-Tech Research Group, recognised since 2022: "Leader in the Global VoC Solution Provider category, 2026". - QKS Group, recognised since 2023: "Leader in the SPARK Matrix™: Voice of the Customer (VoC) Platform, 2026". - Frost & Sullivan, recognised since 2023: "Leader in the Frost Radar™: Voice of Customer Analytics Platforms, 2025" and "European Technology Innovation Leadership Award: Explainable AI for Customer Journeys, 2025". - Everest Group, recognised since 2024: "Key Innovator in Voice of the Customer platforms, 2025". - Analyst quote: "Well positioned as a product leader in the European market with its cutting-edge deep machine learning and LLM technologies." Ankita Singh, Senior Industry Analyst, Frost Radar™: Voice of Customer Analytics Platforms, 2025. ## 16. sandsiv+ compared with Medallia and Qualtrics SANDSIV publishes two comparison pages, in English: https://sandsiv.com/compare/sandsiv-vs-medallia and https://sandsiv.com/compare/sandsiv-vs-qualtrics. SANDSIV builds one of the products compared, so they are a vendor's comparison. Their main points, with the sources the pages name: - **Jurisdiction.** "Medallia is US-headquartered, so a US court can compel it under the CLOUD Act wherever the servers physically stand. A European region changes the geography, not the jurisdiction." "Qualtrics is US-headquartered, so a US court can compel it under the CLOUD Act wherever the servers physically stand." "SANDSIV is Swiss with no US parent." - **Cost, from third-party data.** "SpendHound puts the average enterprise Medallia bill at USD 598,947 a year." "Duke University published its own numbers: licensing up 50 percent in a single year, an opening ask from Qualtrics of six times the previous price, and a renewal landing around 70 percent above the year before." - **Qualtrics' strength, in SANDSIV's words:** "Qualtrics builds good survey research tools. Running a live CX programme on top of one is a different job, and that is where the waiting and the bills start." - **User reviews.** "In the 2026 SoftwareReviews Voice of the Customer category, scored entirely on verified end-user reviews, sandsiv+ places first at 8.9. Medallia Experience Cloud places tenth at 7.5." And: "Qualtrics XM for Customer Experience places ninth at 7.6." These are user-review scores, not a capability audit. - **What sandsiv+ offers instead:** - "Autonomy, in hours. Your team configures journeys, dashboards and surveys. No gatekeeper, no certified administrator, no third party consulting firm." SCOTY builds the survey, the dashboard or the taxonomy when asked. - "One annual fee covering unlimited volume, users, surveys and dashboards. No response buckets. No overage rate above the base price." - "AI you can interrogate. Ask in plain language, get the answer with its working attached: the query, the row counts, how fresh the data was." - A free migration assessment, then one week for the move, with the current platform running alongside (section 13). SANDSIV publishes no comparison with other vendors. For any other platform, the migration tooling still applies (section 13), and SANDSIV compares against a buyer's current set-up in a call. ## 17. InsightNarrator or sandsiv+? - **InsightNarrator on its own** fits a team that already collects feedback and wants to analyse it: upload files or connect survey tools and review sites, with no integration project. "Upload feedback, get the narrative. No integration required." Public prices, free to start, 14-day trial. - **sandsiv+** adds capture on every channel (surveys by email, web, app, SMS, WhatsApp and IVR), contact rules, case routing into ticketing, single sign-on with the customer's identity provider, hosting in the chosen country or the customer's own data centre, unlimited users, migration included and a named expert. - **Moving from one to the other:** "You can start free with InsightNarrator, our agentic analysis module, and expand when you are ready." The free-trial page says: "Scale into sandsiv+ when ready". ## 18. Questions buyers ask, with answers **What is sandsiv+?** SANDSIV's agentic Voice of the Customer platform for enterprises. It captures feedback on every channel, AI agents find the causes, a business case puts a figure in euros on them, and each case goes to the person who owns the fix. SANDSIV is Swiss, with no US parent, and sandsiv+ runs in Germany or Switzerland, or in the customer's own data centre. **What does "agentic" mean here?** The difference is what you get at the end. A dashboard tells you satisfaction fell. In sandsiv+, agents read the feedback, find the cause at aspect level and show the evidence behind it; a business case puts a figure in euros on it from your own numbers; and the case goes to the person who owns the fix. **What is the difference between sandsiv+ and InsightNarrator?** InsightNarrator is the analysis: upload or connect feedback, and AI agents turn it into findings, stories and business cases. It runs on its own with public plans and a free trial. sandsiv+ is the whole platform: capture on every channel, contact rules, InsightNarrator for the analysis, cases routed into ticketing, single sign-on, hosting in the chosen country, and migration included, with a named expert. **How do we know the analysis is right?** "Because you can check it. The computation is visible rather than asserted: you can see how a figure was reached and trace it back to the source rows it came from." Every SCOTY answer shows its source, the SQL it ran, the row counts and the data's freshness, and every agent report ends with an evidence appendix. **Does it read every comment?** Aspect-based sentiment and complaint risk are scored on every comment, up to 100,000 per job. The report-style analyses in the catalogue work from a random sample of up to 500 comments. SCOTY's answers and the agent reports show the queries and row counts behind them. **What are the AI agents?** A team of 10 agents: an orchestrator (NarratorGPT) that plans the run and writes the report, specialists for data, domain context, sentiment, themes, trends, segments and risk, a synthesis analyst, and a quality auditor that challenges the others' work and scores confidence before you see it. You can steer them while they work. **Can it put a figure in euros on a problem?** Yes, with the business case builder: you set the initiative, the expected improvement and your financial figures, and it calculates a three-year return and present value, with a low and a high scenario, in euros. **Can we use our own categories and definitions?** Yes. Start from an industry template or have a taxonomy generated from your data, then edit it. Your metrics, glossary and rules go into a Knowledge Pack that the analysis uses, and the reference documents the model drafts are approved by a person before use. **Which channels can we collect feedback from?** Surveys by email, on your website (pop-up, button or embedded, with a chat mode), in your Android or iOS app, and by SMS, including answers by reply. WhatsApp and automated phone surveys (IVR) are set up per customer. Reviews and other survey tools come in through InsightNarrator. **Can we bring in feedback we already have?** Yes. Upload CSV or Excel files, import on a schedule by SFTP, or use the REST API. InsightNarrator has 26 connectors to survey tools, including Qualtrics, Medallia, SurveyMonkey, Typeform, Google Forms and Microsoft Forms, and pulls reviews from 25 public platforms such as Trustpilot, Google Maps, the App Store and Google Play. **Do you integrate with Salesforce, ServiceNow, Zendesk or another ticket system?** Connections to ticket systems are set up for each customer rather than switched on from a catalogue, so SANDSIV confirms what is possible for yours. sandsiv+ has 146 connectors, and more are added all the time. **Which languages do you support?** Surveys run in more than 30 languages, open answers in other languages are translated through the translation layer, and InsightNarrator analyses feedback in any language as written and writes its reports in 57 languages. Wider language coverage is being built. The application itself is in English. **Do you analyse call-centre calls?** Speech analytics on contact-centre calls is being built. Today, open answers in IVR survey calls are recorded and transcribed. **Does our feedback train shared AI models?** "Your data stays where you put it and never trains a shared model." **Which AI models do you use? Can we switch AI off?** You choose. You can run on a pure EU stack, with models hosted in Europe such as Mistral, or use the latest frontier models, which their providers process outside the EU. The AI layer can also be switched off. SANDSIV sets out the processing chain for your set-up in writing. **Where does our data live, and who can legally reach it?** "Under the US CLOUD Act, an American court can reach any US provider's data, wherever the servers stand. A European data centre changes the geography, not the jurisdiction. SANDSIV is Swiss, with no US parent to compel." sandsiv+ is hosted in Germany or Switzerland, as you choose, or in your own data centre. **Which certifications do you hold?** ISO 27001, 27701, 27017 and 27018, SOC 2 Type II, CSA STAR Level 1, and GDPR-native. The certificate and the Statement of Applicability are available on request; the SOC 2 report is shared under NDA. **Do you support single sign-on?** sandsiv+ supports single sign-on with your own identity provider over SAML 2.0 (for example ADFS or Azure AD), plus an SMS second factor, set up per customer. InsightNarrator on its own uses sign-in with Google or Microsoft accounts. **How is sandsiv+ priced?** One annual fee, built around the modules you need, where it runs and the systems it connects to. Volume and users are not priced separately; SMS, WhatsApp and IVR are priced, as each carries a carrier cost. Prices start at zero with InsightNarrator; a sandsiv+ figure comes from a tailored quote (https://sandsiv.com/pricing). **How much is InsightNarrator?** Starter €69 a month with 5,000 credits, Business €849 a month with 75,000 credits, Enterprise on request. Annual billing gives two months free. Prices exclude VAT. Current prices: https://insightnarrator.com/pricing **Is there a free trial?** InsightNarrator runs free for 14 days at Business level: upload feedback, get the narrative. For sandsiv+, bring one quarter of your feedback to a demo and see what is in it. **How long does switching take?** A free assessment comes first and ends in a written scope. Then the move itself, included in the annual fee, takes one week, with your current platform running alongside so feedback never stops. **Who do we talk to after we sign?** "The person who onboards you is the one who answers in month twelve. One named expert who knows your setup, rather than a ticket queue and a handover." **Who uses sandsiv+?** The homepage names Swisscom and Kantonsspital Winterthur. More stories: https://sandsiv.com/customers **How do you compare with Medallia or Qualtrics?** See section 16 and the two comparison pages. ## 19. Current figures (these replace earlier figures) Earlier versions of this file (30 August 2026) and some sandsiv.com pages used figures that have changed or that describe work being built. Use these: | Topic | Use this | Replaces | |---|---|---| | Agents | A team of 10 agents, one of them an auditor. A run takes minutes. | "Nine agents"; insights "in seconds" | | Analysis types | More than 30 analysis types | "43 analysis types" | | SCOTY tools | Well over a hundred built-in tools | "104 tools", "10 tool packs" | | Connectors | 146 connectors, with more added all the time. Data also comes in by file, SFTP and the REST API; InsightNarrator has 26 survey-tool connectors and 25 review platforms. | "90+ connectors", "148 connectors", "roughly 145 integrated tools" | | Languages | Surveys in more than 30 languages, a translation layer for open answers, reports in 57 languages. Wider coverage is being built. | "250+ languages" | | Speech | IVR open answers recorded and transcribed. Speech analytics on contact-centre calls is being built. | "Speech analytics on every call" | | Figures in euros | A business case builder that works from the customer's own financial inputs | Findings that arrive with euros attached automatically | | MCP | InsightNarrator's assistant can use outside tools through MCP. Outside agents calling sandsiv+ modules through MCP is not available. | "Every module is a tool your own agents can call, direct or through MCPs" | | Customer-managed keys | Being built | "BYOK is supported" | | Pricing | No tiers; sandsiv+ is quoted on requirements | Three tiers including "Sovereign Enterprise" | | Customers | Only the customers named on the homepage | Longer lists of named customers | Statements in earlier versions of this file that are not repeated here should not be relied on. ## 20. Glossary - **Voice of the Customer (VoC):** a programme that collects, analyses and acts on what customers say about a company, in surveys, reviews, calls, chats and tickets. - **Agentic analytics:** analysis carried out by AI agents that plan, query data, check each other and report, rather than by a single prompt or a fixed dashboard. - **Aspect-based sentiment:** sentiment scored per topic within a comment (for example positive about the advisor, negative about the app), rather than one score for the whole comment. - **Complaint risk (C-Risk):** InsightNarrator's score from 0 to 100 for how likely a comment signals escalation or churn, across 28 risk factors in six groups. - **NPS, CSAT, CES:** Net Promoter Score, Customer Satisfaction and Customer Effort Score, the common experience measures. - **Taxonomy:** the customer's own category tree. The model assigns only categories that exist in it. - **Knowledge Pack:** the governed layer of approved metrics, glossary, rules, data traps, example queries and per-dataset reference documents that the agents consult first. - **Provenance record:** the footer on every SCOTY answer showing source tier, confidence, SQL, row counts and data freshness. - **Deep review:** an optional second reviewer that challenges a SCOTY answer before it is shown. - **NarratorGPT:** the orchestrating agent that plans an agent run and writes the report. - **Quality Auditor:** the agent that challenges the other agents' work, scores confidence and flags contradictions. - **SCOTY:** the assistant inside InsightNarrator, with well over a hundred built-in tools. - **Live Agent:** a scheduled agent check whose report goes to Slack, Microsoft Teams, Telegram, Discord or a webhook. - **Business case builder:** the InsightNarrator tool that calculates a three-year, risk-adjusted return and present value in euros from the customer's inputs. - **Close the loop:** routing a case to the person who owns the fix and tracking it to closure. The inner loop is the individual customer; the outer loop is the cause behind many cases. - **DIGI:** sandsiv+'s website and in-app survey widget. - **VOC Act:** SANDSIV's own case management for cases raised from feedback. - **Quarantine rules:** contact rules that stop the same person being surveyed too often. - **US CLOUD Act:** a US law of 2018 under which US authorities can order a US company to hand over data in its possession, custody or control, wherever the data is stored. - **Pure EU stack:** running the AI features on models hosted in Europe, such as Mistral. - **MCP (Model Context Protocol):** a standard way for AI assistants to use outside tools. ## 21. Company, contact and next steps - **Book a demo:** https://sandsiv.com/book-a-demo. "See sandsiv+ on your own feedback. We show you what moves your scores and how every figure was reached." - **Enterprise and regulated environments:** book a demo or write to info@sandsiv.com. Demo and enterprise requests: "We reply within one business day." - **Tailored quote:** https://sandsiv.com/pricing - **Free trial of InsightNarrator:** https://sandsiv.com/free-trial - **Contact form:** https://sandsiv.com/contact. Email: info@sandsiv.com (sales questions, partnerships, press and analysts, everything else). - **Data protection:** privacy@sandsiv.com. Privacy notice: https://sandsiv.com/privacy - **Customer support (existing customers):** https://support.sandsiv.com - **Received a survey?** "Companies use sandsiv+ to send their own surveys. Questions about a survey, or about your answers, go to the company that sent it." - **Careers:** https://sandsiv.com/careers. Open roles: Senior Account Executive for Italy, Germany and Spain. - **Addresses:** SandSIV Schweiz AG, Technoparkstrasse 1, 8005 Zürich, Switzerland · SANDSIV UK Ltd, Forbury Square, Reading RG1 3EU, United Kingdom · Calle Colón 10, 46004 Valencia, Spain. - **Social:** LinkedIn https://www.linkedin.com/company/sandsiv/ · X https://x.com/sandsiv · YouTube https://www.youtube.com/channel/UCwSNsY41lFrRCaLbAn9OT8w ## 22. Pages on sandsiv.com Product: - [Home](https://sandsiv.com/): agentic Voice of the Customer for enterprises; AI agents, tools, customers, analysts, customer success, where data lives, pricing and migration in short. - [Platform](https://sandsiv.com/platform): sandsiv+ module by module, how it listens, analyses and closes the loop. - [Listen](https://sandsiv.com/platform/listen): the channels sandsiv+ captures feedback on (email, SMS, WhatsApp, IVR, web and app), plus reviews, other survey tools and files, and the contact rules. WhatsApp surveys (the customer's approved templates through a WhatsApp Business provider, answered by typing or with button and list replies; set up per customer) and IVR surveys (automated survey calls, with choices on the keypad and open answers spoken and transcribed) are explained on the Listen page. - [Analyse & Quantify](https://sandsiv.com/platform/analyse): InsightNarrator in sandsiv+: aspects and sentiment, risk, drivers and root cause, the agent team and its auditor, and business cases in euros. - [Close the loop](https://sandsiv.com/platform/close-the-loop): real-time alerts, case management with a named owner, closure from your CRM, and root cause analysis. - [Agents](https://sandsiv.com/platform/agents): the AI agents that analyse customer feedback, quantify the drivers behind NPS and CSAT and route each finding to the team that owns it. - [Connectors](https://sandsiv.com/platform/connectors): the connectors to survey, contact centre, CRM and data tools, and the MCP server. - [Migration](https://sandsiv.com/platform/migration): moving response history, surveys, dashboards and integrations to sandsiv+, included in the annual fee. Pricing and getting started: - [Pricing](https://sandsiv.com/pricing): how the price is built: the modules and your migration, data residency and organisation preferences, for a tailored quote. - [Book a demo](https://sandsiv.com/book-a-demo): a demo on your own feedback with an advisor. - [Free trial](https://sandsiv.com/free-trial): the 14-day trial, which runs in InsightNarrator. - [InsightNarrator](https://insightnarrator.com) and [InsightNarrator pricing](https://insightnarrator.com/pricing): the standalone analysis product. Sovereignty: - [Sovereignty](https://sandsiv.com/sovereignty): SANDSIV is a Swiss company; the CLOUD Act argument; where data lives. - [Data hosting](https://sandsiv.com/sovereignty/data-hosting): the hosting choices (Germany, Switzerland or your own data centre) and where AI processing runs. - [Subprocessors](https://sandsiv.com/sovereignty/subprocessors): who processes data for sandsiv+ and InsightNarrator, and where. - [Certifications](https://sandsiv.com/sovereignty/certifications): what SANDSIV is certified for, with the documents on request. Customers and comparisons: - [Customers](https://sandsiv.com/customers): customer stories and how users rate sandsiv+. - [Customer success](https://sandsiv.com/customer-success): a team of experts who know your business, from the first day of implementation to scaling. - [Analyst recognition](https://sandsiv.com/recognitions): the analyst reports and user-review rankings that recognise sandsiv+. - [Swisscom story](https://sandsiv.com/customers/case-swisscom), [Kantonsspital Winterthur story](https://sandsiv.com/customers/case-ksw) and [Salt story](https://sandsiv.com/customers/case-salt). - [sandsiv+ vs Medallia](https://sandsiv.com/compare/sandsiv-vs-medallia) and [sandsiv+ vs Qualtrics](https://sandsiv.com/compare/sandsiv-vs-qualtrics): jurisdiction, cost, analysis and the migration path, in English. Writing: - [Blog](https://sandsiv.com/blog): notes on Voice of the Customer, agents, and what SANDSIV is building in Zürich. English only. - [What is a customer intelligence agent, and why CX dashboards are the next thing AI replaces](https://sandsiv.com/blog/what-is-a-customer-intelligence-agent) - [Anthropic is right: Agentic Analytics is not text-to-SQL. Here's how we built the stack around it.](https://sandsiv.com/blog/agentic-analytics-is-not-text-to-sql) Company and legal: - [Contact](https://sandsiv.com/contact) · [Careers](https://sandsiv.com/careers) · [Impressum](https://sandsiv.com/impressum) · [Privacy notice](https://sandsiv.com/privacy) · [Cookie policy](https://sandsiv.com/cookies) · [Terms of use](https://sandsiv.com/terms-of-use) · [General terms and conditions](https://sandsiv.com/general-terms-and-conditions) ## 23. Sources - SANDSIV's own published material: sandsiv.com (homepage, platform, Listen, Analyse & Quantify, pricing, sovereignty, data hosting, subprocessors, certifications, customers, comparison pages, free trial, book a demo, contact, careers, impressum, blog) and insightnarrator.com (pricing). - Capability status (available, set up per customer, being built): SANDSIV's product documentation, checked against the product on 28 September 2026. - Third-party figures, as cited on SANDSIV's pages: Info-Tech Research Group SoftwareReviews (Voice of the Customer category, 2026), Frost & Sullivan, QKS Group, Everest Group, SpendHound (Medallia spend data) and Duke University (Qualtrics pricing). These remain those parties' data. Last updated: 4 October 2026.