Cogliva — AI-enabled business strategy workspaceCogliva
Trust & Responsible AI

AI support for better decisions, not a replacement for judgment

Cogliva is built to help consultants and leaders think more clearly. Here is how we approach responsible use, data handling, and the limits of what the platform should do.

Cogliva is decision-support software

It helps you think more clearly and produce better-structured work. It is a tool to support decisions, not an authority that makes them for you.

You remain responsible for outputs

Users are responsible for reviewing, adapting, and validating everything Cogliva produces before relying on it or sharing it with clients or stakeholders.

AI outputs may need interpretation

AI-generated content can be incomplete or require human interpretation. Treat each output as a structured draft to refine with your own knowledge.

Handle sensitive information carefully

Sensitive client and business information should be handled with care. Share only what is appropriate, and follow your own confidentiality obligations.

Responsible use of Strategic Signals

Strategic Signals is structured monitoring and decision support. It does not monitor every source or guarantee complete market intelligence.

Liva assists, you decide

Liva, Cogliva's built-in AI assistant, can draft, refine, and challenge content inline — but its suggestions are never applied until you choose to insert, replace, or append them. Review each suggestion against your own knowledge before acting on it.

Responsible use of Systems content

Systems and standards-related content supports business thinking and preparation for professional review — it is not an audit or certification decision.

Secure access by design

Every account is protected by two-factor authentication, with one-time recovery codes for backup — so your strategic work stays behind more than a password, including for Google and Apple sign-in. Sessions also sign out automatically after a period of inactivity.

Exposure Rule for AI knowledge

Liva and the Management Copilot only draw on knowledge your admins have explicitly marked AI-visible. Files filed under Core files, Archive, System files, or Configuration prompts are never exposed to the AI, and internal source filenames are sanitised from every AI response so raw file references cannot leak into a report.

Industry-aware, freshness-aware

When your organization profile carries a sector, Cogliva prioritises industry-specific guidance and evidence when Liva and the Management Copilot reason about your work. Where guidance leans on time-sensitive market evidence, Cogliva flags freshness so time-anchored data is treated as an input, not a permanent truth.

Voice dictation without audio retention

You can dictate directly into Ask Liva. Audio is used only to produce the transcript in the moment and is not retained. The transcript then becomes user content you review and edit like any other draft — no different from typing it yourself.

Your feedback is not used to train models

Helpful / Not helpful signals on Copilot reports and Ask Liva suggestions are used only by the Cogliva team to improve prompts and safeguards. They are not sent to the underlying model providers for training.

The Cogliva Semantic Control Layer

Due diligence built into how AI is used

Cogliva runs a set of behind-the-scenes safeguards — what we call the Semantic Control Layer — to keep AI output grounded, meaningful, and traceable. It reduces the risk of drifting or made-up conclusions and makes it easier to see how a result was produced. It is designed to improve accuracy and traceability; it does not guarantee correctness, and it never replaces your review.

Grounded in real context, not invention

Cogliva's AI works from your own business context and a curated knowledge base rather than free-form guessing. Analyses are generated against the specifics of the organisation you are working on, so outputs stay tied to real inputs.

Field-level meaning and reuse rules

Behind the scenes, each kind of information carries a defined meaning and rules for how it may be reused — so a regulatory note is not silently turned into a confirmed risk, a stated challenge into a proven root cause, or a recommendation into a committed action. These rules are written into the AI's instructions before it generates.

Automated review checks

After generation, Cogliva runs automated checks that flag questionable transformations for internal quality monitoring. They are advisory signals used to keep improving prompts and safeguards — not a claim that every output is correct.

Source verification for Strategic Signals

For Strategic Signals, headlines are read verbatim from the real page, links are verified, and each claim is checked against what the source actually says. When a source cannot be confirmed, the alert is labelled unverified rather than asserted as fact.

Provenance you can trace

For generated values, Cogliva records how they came to exist — which model produced them and which knowledge sources, frameworks, and context shaped them. This provenance is used by the Cogliva team as an internal quality-assurance tool to inspect and improve how outputs are produced.

You stay in the loop

None of this removes your judgment. Suggestions are only applied when you choose to, and you remain responsible for reviewing and validating outputs before relying on them.

These safeguards are internal engineering measures, not a certification or independent audit. They are designed to improve results — not to promise a specific outcome.

What Cogliva does not do

Cogliva supports professional work — it does not stand in for it.

Replace legal advice
Replace audit or certification
Replace consulting judgment
Replace executive accountability
Replace professional review

Privacy and data handling

We aim to follow privacy-by-design principles in how data is collected, used, and protected. For the full detail, read our Privacy Policy and Terms of Service.

Confidence comes from clarity

See how Cogliva structures the work you stay responsible for.