alpha

One governed source for every AI assistant.

One brain to rule them aLL(M). — EU-native governance layer for AI assistants. Teach your company to AI once — Cursor, Copilot, JetBrains and Claude all work from the same encrypted rules and knowledge, each person's assistant sees only what their role allows.

Request an enterprise pilotSee the problem we solve

brain · context · tools

TAIP architectureKnowledge, rules and tools are served from taip to Cursor, Claude and Copilot, role-filtered before decryption.KnowledgeDocs, wikis, code,templatesRulesPolicies, standards,guardrailsToolsCurated catalogue,not enforcementrolesencryptionauditCursorconnectedClaudeconnectedCopilotconnected

The taip rule library: 475 rules across 19 domains (taip-brain, 2026-07) — built from daily AI-assisted work in a regulated MedTech codebase since 2026.

MCP-compatibleEU-hostedRole-aware

Who taip is for.

solo & vibe coders

Security is the missing half of vibe coding.

Free encrypted security rules protect your AI-generated code from day one — one short MCP configuration, no per-repository rule setup.

For solo developers →

engineering & management

Stop re-explaining your company to every AI.

One encrypted source of truth. Every assistant aligned. Faster onboarding, no per-repo rule drift.

For engineering teams →

EU companies

Governance controls for regulated EU teams.

For organisations in regulated AI use cases: taip's context-mcp records who read what. When a data subject asks to be erased, taip-managed content is crypto-shredded. EU-hosted, under EU law.

For EU companies →

the problem

Your assistants should not invent their own version of your company.

An LLM without context is a wild horse. taip is the blinders and the saddle — direction and control, not a leash. Your rules and IP live as plaintext files anyone with repo access can copy, and every assistant invents its own version of your company.

taip gives every assistant the same encrypted rules and context, filtered to the role that is asking. One source. Role-aware access.

Before and after TAIPBefore TAIP, Cursor, Claude and Copilot each use a different source. After TAIP, all three assistants connect to one governed layer of approved knowledge, policies and tools.before: fragmentedafter: alignedCursorClaudeCopilotTeam wikiLocal rulesOld docsDifferent sources. Different answers.CursorClaudeCopilotApproved knowledge · policies · toolsOne governed source. Consistent outcomes.

founder proof

Built by a MedTech founder who was tired of re-explaining MDR to every assistant.

the founder, Taip Systems OÜ

About taip →

trust & safety

The controls your auditor will ask about.

Encrypted until the moment of use
AES-256 encryption at rest. Decrypted only in memory, after the request is authorized by role.
Everyone sees only their share
Role filtering runs before decryption: authorization checks plaintext frontmatter and fails closed. Identity is established through tenant-level access control. A role reads only what its profile permits.
Your data stays under EU law
Each customer gets its own tenant, hosted in the EU and operated under EU law.
context-mcp: access log for every read and write
context-mcp records every read and write with actor, role, item, and timestamp. Brain-side access logging is planned — until it ships, brain sessions are not individually logged.

Read the security overview →

questions

Good questions.

Is this just .cursorrules with extra steps?

One encrypted set instead of a copy per repo. Roles let you control who reads what. context-mcp records every read. .cursorrules has none of that.

Do you sit between my AI and my tools?

No. taip serves and manages access to what an assistant reads from taip. It is not a gateway or firewall over your other tools and does not intercept or block their calls.

What does alpha mean for our data?

You can export your data and leave at any time. That offer is confirmed and will ship. We onboard companies one at a time so we can make the product right for each one.

What happens when someone leaves?

Revoking the key ends access to TAIP-managed content. A credential held at the time of revocation remains valid until its expiry window — the key governs what taip stores, not an active session. No plaintext IP remains in taip's own storage — there is nothing to revoke in taip except the key itself. Data previously sent to connected AI services must be managed under those services' own deletion policies.

We are six developers, not an enterprise — is this for us?

Yes. The product is built for the 5-to-50-developer EU team and the solo developer. If you are an enterprise with thousands of seats, we are not the right fit yet.