Clara

The AI consultant who runs your transformation project with you.

From a two-line brief to governed, AI-ready knowledge. Six steps, one companion, fully customizable to the way your team works.

Read the docs
The platform

Six steps, one live knowledge base.

What Clara does for your project, screen by screen. Everything she does is yours to customize: her skills, her interview guides, the artifacts she produces and the templates she starts from.

Clarifeye Home: pending interviews, signals and recent changesClarifeye Sources: documents in collections, connectors and external MCPsThe Skills picker in Work with Clara: one-off actions Clara can runA Clara interview in progress, onboarding steps completedWork with Clara beside the Playbooks artifact, with the evidence usedEdit Template: general, scoping, artifacts and skills tabs
01

Two lines in, a plan out.

Tell Clara what the project is about and who runs it. She proposes the artifacts to build and the plan to get there.

  • Brief and Project Plan seeded on day one
  • Artifacts picked from a catalog
  • Home shows what needs you today
02

All the context, in one place.

Upload documents, connect the tools where knowledge already lives, plug in the IT systems through MCP.

  • Documents organized in collections
  • SharePoint, Confluence, Granola connectors
  • External MCPs for your IT systems
03

A junior consultant you fully customize.

Ask in plain language, or pick a skill: analyze interviews, find gaps versus sources, draft the next interviews. Her skills, her interview guides and the artifacts she produces are yours to shape.

  • Skills: analyze, draft, compare, prep, organize
  • Reasons on brief, plan, sessions and library
  • Add your own skills, nothing sends until you do
04

Clara asks what nobody wrote down.

She spots the gaps worth exploring, drafts the interviews, and runs them with the right people: async, scheduled, in their language.

  • Every question informed by what is already known
  • Interviewees answer from a link, in any language
  • Pending interviews and reminders, from the Home
05

Every answer lands in the right artifact.

Clara folds interviews and sources back into the mental map, the process blueprint, the rules and the playbooks. Versioned, with the evidence used.

  • Who said what, every change tracked
  • Evidence used stated, no invented branch
  • Playbooks ready to run as agents
06

Do it once, replay it on every project.

Package the scoping questions, the artifacts and the skills into a template. The next project starts from your best practice, not a blank page.

  • Scoping, artifacts and skills per template
  • Consultants start from it in one click
  • Best practice travels across teams and clients
Interviews

Interviews people actually finish.

Clara runs them so nobody has to book a room. You keep control of what is asked, how, and when.

You keep the wheel

The interview guide is yours. Pick the mode, from a quick check to a deep dive. Schedule it, and let the reminders do the chasing.

Nobody re-explains what is known

Every question is informed by what Clara has already read. She goes to the gap, and flags the contradiction when two answers disagree.

People actually answer

Async, in their language, on a phone or a screen, from a shared link. Empathetic enough that people open up instead of ticking boxes.

You see where it stands

One project view: who has answered, who is pending, what came in today. One click to remind.

The interviews dashboard: ten interviews with their status, drafts ready to review, campaigns grouped by topic, and a one-click reminder for whoever has not finished
The artifacts

What Clara builds for you.

A catalog of structured knowledge, curated per store. You define what each artifact is, what it is for and how it consolidates. Clara does the consolidating, on her own, every time something new comes in.

Click an artifact to open it below.

Brief Scope, teams, sites and what the outcome must enable.
Brief artifact: the problem and the context of a self-service contract change initiative, and the specification the project must produce
Project Plan Sources, systems, owners and adoption steps.
Project Plan artifact: goals with their owners, then the interviews to run in each wave and what to obtain from them
Mental Map The operation's shared language across teams and sites.
Mental Map artifact: five colour-coded domains, from workflow steps to regional operating variations, and how their concepts connect
Process Blueprint BPMN 2.0 of how the work is actually performed.
As-is Process Blueprint, a BPMN diagram across five lanes with exception handoffs
Business Rules Every rule with its source, owner, status and exceptions.
Business Rules and Exceptions: rule statement, source, owner, status, exceptions
Playbooks Operating instructions, ready to run as agents.
Playbooks artifact: three operating procedures drawn as step graphs, each step typed Analyze, Retrieve or Format
Glossary One multilingual vocabulary, tied to the Mental Map.
Glossary artifact: process terms in English, German, French and Italian, each with its definition in the client's process
Library Description Which document collections exist and how the process uses them. Or any Markdown file you define.
Library Description artifact: each document collection with its source, what it contains and how the process uses it, including live Jira and Confluence read through MCP

Don't see the artifact you need?

We define it with you: what it is, what it is for, how it consolidates. From then on Clara maintains it automatically, like the rest of the catalog.

Adapts to your project

Built for the way a programme actually runs.

Gaps and contradictions

Surfaced before you interview anyone, not three months in.

Contextualized interviews

Every question informed by what is already known. Async, any language.

Consolidate automatically

Every answer folded back into the outputs, no write-up backlog.

Audit and versioning

Know who said what, every change tracked, rollback anytime.

Collaborative by nature

Built for teams: contribute, review, share.

AI native

Connect it to any AI through MCP, upstream and downstream.

Open by design

Your knowledge stays yours, and stays portable.

What goes into Clarifeye is meant to come back out. Open standards to your AI stack, open formats for every artifact.

Use it in
ClaudeChatGPTMicrosoft CopilotGleanWriterand any MCP client
Connect it to
SharePointGoogle DriveDatabricksSnowflakeSalesforceSAPand any MCP server
Open formats
Markdown, JSON and standard schemas. Every artifact readable and editable without Clarifeye.
MCP and API
Read and write the store from Claude, ChatGPT, Copilot, your own agents, or a REST API for older integrations.
Mirrored where you work
JSON mirror into Snowflake, Databricks or blob storage, kept in sync.
Export anytime
Full reversibility and export at no extra cost. Your knowledge leaves with you.

Safe by default.

Enterprise controls, your data where you decide, and a record of every change. Nothing is used to train models.

  • SOC 2 Type II
  • EU or US residency
  • Zero data training
  • Cloud, Dedicated Cloud or Customer Cloud

Technical FAQ

Clara is the AI consultant who runs the project with you. She reads your documents and systems, works out what is missing, prepares and runs the interviews with the right people, and folds every answer back into the artifacts. She comes with skills, interview guides and a continuous discovery loop you can shape to your method. The knowledge stays entirely yours.

Yes. Clara can interview multiple subject-matter experts, incorporate conversations across teams, and reconcile different points of view into a shared, versioned knowledge layer. Everyone contributes. Every change is tracked.

Your knowledge lives in open formats: Markdown, JSON, and standard schemas. Briefs, playbooks, corpus, mental maps, all readable without Clarifeye. Explicit and implicit knowledge are versioned and linked. You can leave at any time and take everything with you.

Yes. Every change is tracked. Knowledge data, skills, and tools are fully versioned. Teams can roll back at any time, and choose between auto-evolution for fast-moving use cases or manual approval gates for critical ones. Every change is validated against prior interactions before promotion, so regressions are caught before users see them.

Yes: images, flowcharts, diagrams, tabular content. PDF, PPT, .md, .docx, scanned or not. We continuously improve parsing capabilities and use specialist enrichment techniques to contextualize data reading based on knowledge already captured by Clara.

EU and US data localisation. SOC 2 Type II. Zero training on customer data. PII detection and redaction available at the ingestion layer. A full list of subprocessors is available in our DPA (available upon request). For sensitive workloads, Enterprise adds Dedicated Cloud or Customer Cloud deployment.

No. Your knowledge is never used to train models. Ever. Your IP stays your IP, on every plan.

Three options, in increasing order of isolation: Cloud, our multi-tenant platform and the default; Dedicated Cloud, a single-tenant instance we run for you; and Customer Cloud, Clarifeye running inside your own cloud account, still fully managed by us. Dedicated Cloud and Customer Cloud are Enterprise options. Whichever you pick, the platform behaves the same; only where it runs changes.

Clarifeye is compatible with all major providers: Anthropic (Claude), OpenAI (GPT), Google (Gemini), and Mistral. Logic is portable, swap models without rebuilding. BYOK options are available on the Enterprise plan.

Teams do, and it holds for a quarter. Then it drifts: the wiki says one thing, the people another, and nobody owns the reconciliation. Tuning never ends and it is not the transformation team's job. Clarifeye keeps the base current because capture is continuous: every new interview, signal or source folds back in, with the contradiction flagged and the evidence attached, and Clara proposes the next step instead of waiting for someone to notice.

Want to dig deeper?

Walk through integration patterns, deployment options and security with our solutions team, or read the engineering docs first.

Read the docs