Studio
Black Friday

Up to 70% off kits30% off services

Ends in
All Tools
AI agents · Free assessment

AI Agent Readiness Assessment

Pick one task you'd like an AI agent to take over, then answer fifteen yes-or-no questions about the job, your data, your tools, risk and ownership. Five minutes, a readiness score, and what to fix before anyone builds anything. Answers stay in your browser.

It appears at the top of your copied results, so you can compare several tasks.

1.The job

  1. Can you describe the task in one sentence: what comes in, what gets decided, what happens next?
  2. Does the task come up at least 200 times a month?
  3. Does the task need someone to read or judge messy input, like emails, PDFs or free text?
  4. Would two of your team members usually give the same answer for the same item?

2.Your data

  1. Can you export at least 50 real past examples of the input, with what the right outcome was?
  2. Do the answers the agent needs live in one place, like saved replies, a price list or a knowledge base?
  3. Do you know what data may be sent to an AI model provider, and what must stay in-house?

3.Your tools

  1. Do the tools the agent reads from and writes to have an API or a supported integration?
  2. Is the data in those tools reasonably clean, without lots of duplicates or stale records?
  3. Do you have, or can you open, your own account with a model provider such as OpenAI, Anthropic or Google?

4.Risk and control

  1. Have you decided which actions need a person to approve them, such as emailing customers or moving money?
  2. Do you know what a wrong answer costs, and how you would notice one?
  3. Will every decision the agent makes be logged with the input and the reason?

5.Ownership

  1. Is there one named person who owns the agent after launch?
  2. Have you agreed how success will be measured, such as accuracy, time saved or tickets handled?

Want it tested on your real data?30% OFF

Our $349$499 opportunity assessment scores up to 5 tasks and tests the best one on 50 of your past examples with two models: accuracy, cost per run and payback in about 5 business days, credited toward the build.

Why most AI agent pilots stall, and how to avoid it

Agent projects rarely fail because the model isn't clever enough. They stall because the job was vague, nobody had examples to test on, the tools couldn't be connected, or nobody decided what the agent was allowed to do alone.

  1. Pick one job, write it in one sentence. What comes in, what gets decided, what happens next. "Use AI for support" is a wish; "tag and route every new ticket" is a job.
  2. Collect 50 real examples with the right answers. It's the only way to know the accuracy before go-live instead of after a customer complains.
  3. Check the money. Our ROI calculator shows the payback, and the token cost calculator prices the model usage on 9 models.
  4. Start in suggest mode. The agent drafts; a person approves. Switch single steps to automatic only when the decision log shows they're right.
  5. Name an owner. Models, prompts and your tools change. Someone has to read the log and rerun the tests.

Frequently asked questions

What the assessment covers, how it scores and what to do next.