TellseedRequest access
How it worksUse casesMCPPrivate betaRequest access

Private beta

Give your AI better human data.

Turn what your AI gets wrong into a focused campaign for human knowledge. Create it through the agent you already use. Collect examples from the people who know your domain.

Request private beta accessSee how it works
A starting point

“Our support assistant keeps confusing cancellations with refunds.”

An illustrative case, followed through this page.

Tell your agent what your AI needs to learn.

How it works

From a known weakness to usable human data.

Follow one support-routing problem from a conversation to a reviewed example.

Worked example Support routing

  1. Your agent
  2. Tellseed MCP
  3. Campaign
  4. Domain experts
  5. Reviewed data

  1. Tell your agent what needs improving

    Explain the actual problem in natural language: routing failures, extraction mistakes, missing domain knowledge, weak evaluation coverage.

    Your agent asks
    I can create a Tellseed collection campaign for that. Who should contribute?
    You decide
    “Our support specialists — the people who actually resolve these cases.”

  2. Tellseed prepares the campaign

    Your agent uses Tellseed through MCP to draft the objective, scenarios, contributor instructions, target structure, branding and collection settings.

    The campaign brief
    Objective
    Improve refund vs cancellation routing
    Contributors
    Support specialists
    Coverage to collect
    16 scenarios

    3 decisions needed

    Review the draft and resolve the missing decisions before approving publication.

  3. Experts contribute

    Contributors open a normal web page. No MCP, no AI account, no technical setup — they just provide the input you asked for.

    An expert contribution

    “Please stop my subscription after this month. I don’t need any money back.”

    Intent
    Cancellation
    Expert reasoning
    The customer wants to stop future renewals. They explicitly say they are not asking for a refund.
    Illustrative contribution and expert reasoning.

  4. Bring the data back

    Track coverage, review contributions, and retrieve structured datasets for AI development.

    A reviewed example

    Human input
    Please stop my subscription after this month. I don’t need any money back.
    Intent
    Cancellation
    Expert reasoning
    The customer wants to stop future renewals. They explicitly say they are not asking for a refund.
    Source
    Support specialist contribution
    Scenario
    Cancellation without a refund request
    Review state
    Reviewed
    Illustrative data view; not a published export schema.

Where the work goes.

The context starts with your agent and stays with the collection.

A typical workflow

  1. Find a tool
  2. Configure a form
  3. Explain the task again
  4. Send the link
  5. Manually repair the output

With Tellseed

  1. Tell your agent
  2. Approve the campaign
  3. Send the link
  4. Retrieve structured data

Use cases

What teams collect with Tellseed.

Tellseed is designed for campaigns where the missing ingredient is human judgement, not more scraped text.

  • Intent and routing

    Collect how real users express requests that should map to categories or actions.

  • Structured extraction

    Gather natural questions that should resolve into structured fields or semantic frames.

  • Domain expert knowledge

    Collect expert explanations, decisions, corrections, examples, and edge cases.

  • Evaluation datasets

    Create focused human-written tests around known model weaknesses.

  • Failure-driven campaigns

    Turn recurring AI failures into targeted rounds of fresh human examples.

MCP Technical reference

Built for the agent-first workflow

Campaign creation stays in your existing AI assistant. Tellseed provides the collection workflow through MCP.

Campaign operations

  • Create a campaign draft
  • Update requirements
  • Inspect missing configuration
  • Preview a campaign
  • Publish an approved campaign
  • Check collection status
  • Retrieve exports
Illustrative agent sessionmcp · tellseed

Create a Tellseed campaign for the support-routing failures we discussed.

Tellseed

Campaign draft created.

Objective
Improve refund vs cancellation routing
Audience
Support specialists
Scenarios
16 scenarios
Missing decisions
3
Illustrative. Tool names and output shown here are not a published interface contract.

Private beta

Request private beta access

We are working with a small number of AI teams to shape the first version of Tellseed. If you are building or evaluating an AI system and need better human data, tell us what you are working on.

Currently accepting a small number of private beta teams.

What happens next

  1. You describe the AI problem you are trying to solve.
  2. We read it and check it against what the first round is focused on.
  3. If it fits, we get in touch to set you up.

Your beta application

We use what you send to assess fit for the private beta. Nothing else.