Trulyp AI

From questions to answers. From answers to action.

Explore trusted data, get help, and manage your workspace through conversation. Chat in Trulyp or use ChatGPT and Claude to access 300 tools through MCP. Connect other MCP servers to bring more context and capabilities into Trulyp.

Conversational analytics

Ask naturally. Get precise answers.

Explore trusted data in plain language. Trulyp turns your questions into deterministic queries, so answers come from calculated results you can rely on.

Connected through MCP

Trulyp, in your AI assistant.

Chat with your data, navigate your workspace, and administer Trulyp from ChatGPT or Claude. With 300 tools available through MCP, your AI assistant can turn a conversation into action.

Help when you need it

Less searching. More doing.

Ask how something works and get practical guidance grounded in Trulyp’s documentation. From your first connection to advanced workflows, find the next step without leaving your work.

AI Trace

Ask. Then see how it got there.

Vizzy answers with queries on your semantic models, not with guesses. AI Trace records every run step by step, from your question to the exact tool call and query behind the answer.

AI Trace

Live network of AI actions. New tools appear as they are called. Failures show their reason.

Runs

No runs yet. Use the assistant first.

Network

Ask Vizzy a question to see its network.

Tools

No tool activity captured for this run yet.

Node Detail

Select a node to inspect its attached events.

Pick a question for Vizzy. Click any node in the network to inspect it.

How AI tracing keeps every answer auditable

AI agents

Why build your own agents.

Vizzy answers questions as they come. An agent does one job you define, the same way every time, wherever you need it.

  • Automate the work you repeat

    Turn a question or task you handle again and again into a flow that runs the same way every time. Trulyp ships with agents for KPI summaries, cockpit analysis and filling in missing values.

  • Put AI where the work happens

    Call an agent from another AI feature, from a workflow or from your own application, and get a structured answer back.

  • Stay in control

    You choose the model, the tools it may use and the knowledge it can read. Draft, Active and Paused decide when it is available.

  1. 1

    Create

    Give the agent a name, a slug and a description. It starts as a Draft, so nobody can use it yet.

  2. 2

    Build the flow

    Drag blocks onto the Visual builder: an Input payload, AI actions with their prompt and MCP tools, conditions or an intent router, knowledge, data, workflows and a structure for the output.

  3. 3

    Test

    Use Quick test to try a single prompt, or Flow run to run the complete flow and see every step it took.

  4. 4

    Activate and use

    Set the agent to Active and call it from Trulyp, from a workflow or through the API with its slug.

AI you stay in control of

Powerful, and governed by design.

Choose your own model, decide who can use it, and keep every action visible.

  • Bring your own model

    Connect OpenAI, Anthropic, Azure OpenAI or any OpenAI-compatible endpoint, and give each configuration to specific teams.

  • Confirm before it changes anything

    Actions that create, update or delete stop at a Confirm step, so nothing changes without your approval.

  • Tools follow your permissions

    Vizzy only sees the tools, data and rows your role allows, in Trulyp and in ChatGPT or Claude.

  • Insights with evidence

    The insight engine ranks findings, flags anomalies and checks them statistically before Vizzy answers.

  • MCP in both directions

    Use Trulyp from ChatGPT, Claude or Mistral, and connect other MCP servers so Vizzy can use their tools too.

  • Usage you can see

    Track AI token usage per configuration, per user and over time, so costs never come as a surprise.

Read more about AI security in Trulyp