A complete AI governance platform

Real-time DLP for AI prompts

Most AI data leaks happen in the prompts. Employees include customer names, contracts, credentials or medical data in their questions to ChatGPT without realizing the risk.

coordinat.io analyzes every prompt before sending it to the model. If it detects sensitive data, it can block it, warn the user, anonymize the sensitive data or send it for manual review by the security team.

• PII detection — names, emails, ID numbers, IBAN, phone numbers
• Financial data — account numbers, contracts, internal figures
• Credentials — passwords, tokens, API keys
• Health data — diagnoses, medication, medical records (Premium)
• Configurable actions: block, warn, anonymize, review

One entry point for all your AI models

coordinat.io exposes an API that is 100% compatible with the OpenAI specification. Your applications, tools and employees keep working exactly the same — only the base URL changes.

Behind the scenes, coordinat.io handles authentication, applies security policies, logs activity and can route to the most suitable model based on task complexity.

AI Orchestrator — intelligent routing that pays only for what you need

Not every question needs the most powerful model. Asking "what is 15% of €240?" shouldn’t cost the same as "analyze this 40-page contract and extract the risk points".

coordinat.io's AI Orchestrator automatically classifies each request into three complexity levels and sends it to the cheapest model that can solve it correctly. And it's not a static table: every night it re-analyzes real downgrade rates from the last 30 days and reorders which model gets tried first at each tier — the system keeps improving itself, with no configuration needed.

Budgets by organization, department, project or user

AI spend can spike without warning: an employee chains calls to an expensive model, a project goes over budget, an entire department goes unnoticed until the monthly bill arrives.

With coordinat.io you set spend limits at any level — organization, department, project, or individual user — with whatever priority you decide. As spend approaches the limit, the team gets automatic email alerts; once exceeded, the policy can block new requests, force a downgrade to a cheaper model, or simply warn, depending on how you configure it.

• Limits by organization, department, project or user — with configurable priority
• Monthly or total periods, matched to your billing cycle
• Automatic email alerts at multiple thresholds (80%, 90%, 100%)
• On exceeding the limit: block, downgrade to a cheaper model, or just warn
• Each employee sees their own usage and budget status in real time

RAG — The assistant answers from your own documents

AI models know a lot about the world, but they know nothing about your internal processes, company policies or documents. The result: generic answers when an employee asks about something specific to the organization.

With coordinat.io corporate RAG, assistants consult your internal knowledge bases before responding. Upload PDFs, Word, Excel or TXT files and get accurate answers that cite the source.

• Upload internal documents: manuals, policies, contracts, FAQs
• Semantic retrieval via cosine similarity — finds the most relevant chunk
• Context is automatically injected before each response
• DLP applied to content retrieved from documents too
• Link each knowledge base to the assistant that needs it

AI Agents that act — with human control

A chat assistant answers questions. An AI agent can act: search the web, query external APIs, submit forms, update systems. The difference is the ability to execute real tools.

coordinat.io includes an agent registry with authorized tools, a human-in-the-loop approval flow for sensitive actions, and full traceability of every execution. The admin controls what each agent can do.

• Define which tools each agent can use (per-name allowlist)
• Prior approval for actions that require it — human-in-the-loop
• Full execution tree: input, output, duration and cost per step
• Session budget to cap the cost of each agentic run
• DLP applied to every tool call before it executes

Slack, Teams & Jira — where your team already works

Asking employees to switch tools to access corporate AI reduces adoption. coordinat.io integrates with the most-used platforms: the assistant reaches your team where they already are.

Every message arriving via Slack, Teams or Jira goes through the same gateway — with DLP, policies, budget tracking and full logging. A single control point for all channels.

• Slack: slash commands & Events API with HMAC-SHA256 signature verification
• Microsoft Teams: outgoing webhook with HMAC verification
• Jira Automation: inbound webhook with shared secret + ADF comment reply
• Each integration uses the assistant and model you configure
• Full log of every interaction in the audit panel