---
title: "Tanso: find and fix what stops agents from becoming customers"
description: "Tanso helps B2B software teams improve how AI agents discover, access, buy, and use their products through evaluations, implementation help, and retests."
date: 2026-09-05
canonical: https://tansohq.com/
---

# Find and fix what stops agents from becoming customers.

Tanso helps B2B software teams make their products easier for AI agents to
discover, understand, access, buy, and use. We evaluate real customer tasks,
help your team implement improvements, and retest, with clear permissions
and human handoffs where needed.

- [Request an evaluation](https://tansohq.com/#request-evaluation)
- [See an example evaluation](https://tansohq.com/evals#example-evaluation)
- [Talk to a founder](https://cal.com/katrina-laszlo/30-minute-meeting)
- [Full machine-readable context](https://tansohq.com/llms.txt)

The request form requires only a work email. Product website and the task to
evaluate are optional. A plain domain such as example.com is accepted. A submission requests a conversation about an evaluation;
it does not create an account or complete a purchase. Agent Readiness pricing
has not yet been published. There is no supported public signup, payment,
waitlist, or evaluation API for this service.

## Who Agent Readiness is for

Agent Readiness is for product leaders at B2B software companies. The initial
fit is developer tools and API-first SaaS, where agents need to evaluate,
access, and use the product on a customer's behalf.

The evaluation follows the way you sell. A self-serve task might involve
choosing a plan, signing up, and making a first API call. A sales-led task
might involve qualifying the product, making the intended human handoff, and
using granted access. An intended human handoff is part of the task, not an
automatic failure.

## What we evaluate

We follow the customer task across the interfaces your product provides,
including its website, documentation, APIs, and tools.

- **Find and understand:** Can agents identify your capabilities, understand pricing, and find reliable documentation?
- **Access and try:** Can they start a trial, get scoped access, and continue through an intended human handoff?
- **Use and transact:** Can they use the right tools, complete the task, and make or hand off a purchase?
- **Stay in control:** Can people set limits, review consequential actions, and verify outcomes or recover from failures?

These areas summarize the scope; the report retains the seven checkpoints
listed below. The evaluation follows the journey and permissions agreed with
your team.

### Tools we use in this work

Katrina Laszlo built these tools to inspect agent readiness and identify
implementation work:

- [aeo-ready](https://github.com/katrinalaszlo/aeo-ready): combines agent-readiness checks and prioritizes the findings. It produces recommendations; it does not patch the site itself.
- [agent-serve](https://github.com/katrinalaszlo/agent-serve): a skill for auditing onboarding, authentication, purchasing, usage, account management, and developer readiness.

Run the scanner against your own site:

```bash
npx aeo-ready scan https://yoursite.com
```

Install the agent-serve skill:

```bash
npx skills add katrinalaszlo/agent-serve
```

### Research behind our approach

Founder Katrina Laszlo writes about the design and implementation of agent
experiences:

- [Agent Experience](https://katrinalaszlo.com/notebook/agent-experience.html): designing products agents can understand and use.
- [Making Your Product Agent Self-Serve](https://katrinalaszlo.com/notebook/agent-self-serve.html): the path from discovery through access, integration, and payment.

## What you get

- **Recorded agent task runs:** see the steps the agent took and where it got stuck.
- **Findings report:** evidence, the effect on the customer task, and prioritized fixes.
- **Implementation help:** work with Tanso to address product, documentation, and interface blockers.
- **Retests:** run the same task under the agreed conditions to check whether a fix worked.

## Example evaluation

The example at [See an example evaluation](https://tansohq.com/evals#example-evaluation)
is illustrative. It is not a recorded customer run or a claim about measured
performance.

**Example task:** Evaluate an API product, create a trial account, and make a
first API call. A signup step that requires human inbox verification without
a supported handoff can block an agent before it gets credentials. The finding
should identify that step, show the evidence, and recommend delegated
verification or a supported handoff that preserves the required verification.
Retest the same task after the change.

## The seven checkpoints

Every customer journey uses the same checkpoint vocabulary:

1. **Discover:** Can an agent find and retrieve the right facts?
2. **Understand:** Can it tell what the product does and who it is for?
3. **Try:** Can it test value before purchase in a sandbox or free path?
4. **Sign up:** Can it create an account without a human-only gate?
5. **Access:** Can it get scoped credentials and understand their permissions?
6. **Pay:** Can it complete or safely hand off a purchase and confirm the result?
7. **Use:** Can it finish the task it came to do?

The evaluation is scoped to the way the company actually sells. A demo request
is not a failure when it is the intended next step; an opaque or human-only
handoff is.

## What a report contains

Each report row uses this schema:

- `checkpoint`: `Discover`, `Understand`, `Try`, `Sign up`, `Access`, `Pay`, or `Use`
- `status`: `ready`, `partial`, or `blocked`
- `evidence`: a source URL, request and response, or direct observation
- `blocker`: a specific blocker or `null`
- `recommended_next_step`: the highest-priority fix or verification step

Reports include the task definition, starting conditions, allowed tools,
agent/model, timestamp, success criteria, and supporting evidence. These
conditions make a retest comparable. Findings describe the observed run;
controlled evaluations do not reconstruct every journey happening in
production or guarantee success for every agent.

## Agent Readiness Evals

[Agent Readiness Evals](https://tansohq.com/evals.md) turns a real buying or
product task into a repeatable controlled evaluation. Teams can compare a
documentation, interface, prompt, model, checkout, or workflow change under the
same success criteria.

## How it works

1. **Agree on the task:** define the customer goal, allowed access, and what success looks like.
2. **Run it and record the findings:** get a recorded agent run and a report connecting each blocker to evidence.
3. **Fix and retest:** work through the fixes with Tanso, then repeat the task to see what improved.

## Agent Monetization

We built and open-sourced [Agent Monetization](https://tansohq.com/product/agent-monetization.md),
one self-hosted engine for customer billing and internal AI spend. Its scoped
access and agent payment flows inform our work on agent experiences.
Use customer monetization, internal spend, or both in the same installation.
It is free to self-host under AGPL-3.0.

- **Customer monetization:** usage billing, credit wallets, and margin per customer.
- **Internal AI spend:** track provider costs, allocate spend to teams, and set budgets.
- **Agent access and payments:** optional programmatic signup, scoped keys, and purchase flows.

- [Agent Monetization source](https://github.com/tansohq/tanso-oss)
- [Agent Monetization documentation](https://tanso.mintlify.app)
- [Agent integration guide](https://tansohq.com/AGENTS.md)

## Tools

- [AI Pricing Finder](https://tansohq.com/ai-pricing-finder.md): compare pricing,
  credits, commitments, rollover, and overages with source evidence.
- [Margin Analyzer](https://tansohq.com/margin-analyzer.md): identify profitable
  customers and workloads; CSV data stays in the browser.
- [Credit Estimator](https://tansohq.com/credit-estimator.md): model credit burn,
  depletion dates, and purchase plans.

## Tanso blog

The [Tanso blog](https://tansohq.com/blog.md) covers agent readiness, agent
experience, agent monetization, and AI economics. It is separate from Katrina
Laszlo's personal blog.

## Machine-readable links

- [Agent Readiness overview](https://tansohq.com/index.md)
- [Agent Readiness Evals](https://tansohq.com/evals.md)
- [Agent guide](https://tansohq.com/AGENTS.md)
- [Full LLM context](https://tansohq.com/llms.txt)
- [Agent Monetization source](https://github.com/tansohq/tanso-oss)
- [Sitemap](https://tansohq.com/sitemap.md)