Skip to content
Usage analytics

AI agent analytics that show what your agents deliver, and what it costs

See how many Tasks your agents complete each day and how many succeed. Follow any Task step by step, and read model spend by provider and by model. Then choose your next model or use case from your own results.
See Task pricing

Questions your weekly review can answer

The answers come from the Tasks your agents already run, with nothing to add to your applications.

  • What did your agents deliver?

    Completed and successful Tasks for each day, across your organization or for one agent.

  • How did a Task reach its result?

    Every step of a Task, with the models and Tools it used, its duration and its model cost.

  • Which Tools can your agents count on?

    The success rate of each Tool and each MCP server your agents call.

  • What does each model cost?

    Model spend by provider and by model. The Duale AI price of each Task stays fixed.

  • Which model fits the work?

    LLM Coach compares the models in your pool on the skill the work needs, on time per Task and on cost.

  • Who can read the figures?

    You grant read access by permission, so finance, team leads and reviewers see the same figures.

Open any Task and see where its time and spend went

When a team asks how a result was produced, the Task shows its path step by step.

Spot the step that failed, ran long or cost the most, down to the Tool it called. See how you connect your systems.

Choose models with your own evidence

Your Tasks show where another model would help. LLM Coach puts the candidates side by side.

Enable the candidate that fits and follow next week’s success and spend in the same views. Model routing then runs each part of the work on a model you enabled. See the evidence behind each skill.

Read each cost on its own line

Budget the Duale AI line in advance, and follow model spend Task by Task.

Your Duale AI line is the number of Tasks times their fixed price, standard or fast. Your providers invoice model spend directly, and Usage analytics breaks it down by provider and by model. See Task pricing.

Decide the next rollout from one Monday review

An illustrative example: your platform team checks last week's after-call reviews before a second team starts.
From delegated work to the next rollout decisionYour teams delegate Tasks. Your agents deliver results. Usage analytics shows what the Tasks delivered and what they cost. You choose the next models and use cases, and your teams delegate the next Tasks.

The platform team runs after-call reviews for customer service, one Task per technical support call. It plans to open them to the sales team. On Monday morning, Usage analytics shows:

  1. Tasks. The after-call review agent completed 4,200 Tasks last week. Successful Tasks dipped on Thursday.
  2. One Task. A Thursday Task shows the failed step: filing the call wrap-up through the ticketing Tool.
  3. Tools. The ticketing MCP server’s success rate fell during Thursday’s maintenance and recovered on Friday.
  4. Cost. Each Task shows its model spend, by provider and by model.
  5. LLM Coach. A faster model in the pool scores as well on the skill the work needs, so the team enables it.

Decision. On Tuesday, the sales team gets its own review agent with its own quality grid. Next Monday’s review compares both agents.

See what each use case delivers

Usage analytics follows the work each team hands over.

  • Sales

    See how many questionnaire Tasks succeed each week, and open any of them to read its path.

  • Due diligence

    Follow each due diligence Task branch by branch in the Dashboard Task view, and read what each model cost.

  • Customer service

    Count the after-call reviews your agents complete each day, and read each one’s model cost, before you open the work to another team.

  • IT and security

    Follow each alert’s Task step by step, from the Tools each branch called to its model cost.

Before you read your first figures

  • One fixed price per Task

    A standard Task costs €0.10 and a fast Task €0.20, excluding VAT. See pricing

  • Your providers, your contracts

    Your model providers bill you directly for inference.

  • Unlimited agents

    Add agents as your teams and use cases grow.

Usage analytics questions

From your agents’ first Task, so the first weekly review of a pilot already covers every Task it ran and what its models cost. Choose the use case for your pilot.

No. Usage analytics is ready in the Dashboard, so a business sponsor reads each week’s results without waiting for a report. See how you grant read access in Identity and access.

Yes. The figures for each Tool and each MCP server show which agents call it. See how you choose which agents reach each system in Tools.

Yes. Only the people you authorize open each team’s Tasks in detail. See how each person gets only the access their job needs in Identity and access.

Plan your first use case and follow its results from the first Task.

See Task pricing