StewAI

What is StewAI?

StewAI is a platform of tested AI workflows called recipes. Each recipe fixes its inputs, steps, models, checks, and outputs in advance. You can run recipes from the marketplace in your browser, or let Claude, ChatGPT, Cursor, or Codex call them through a plugin. You can also build your own.

Top Features:

• Marketplace: browse hundreds of ready-to-run recipes for research and review work.
• Agent plugin: connect Claude or ChatGPT with one email sign-in, no API key.
• Visual builder: chain AI, search, loop, and export steps without writing code.

Use Cases:

• Survey testing: simulate how customer segments answer a questionnaire before it goes out.
• Contract review: check two contract versions for terms that quietly changed or vanished.
• Data extraction: pull structured fields from documents and export them as CSV.

Who Can Use StewAI?

• Analysts: run repeatable research and forecast checks with one fixed method.
• Legal and procurement: review contracts, policies, and tender bids with structured, checkable outputs.
• Agent users: hand demanding subtasks to tested recipes and save agent tokens.

Pricing

• Free ($0): 500 credits each month, enough for about five small runs.
• Basic ($10/month): 60,000 monthly credits for hundreds of recipe runs each month.
• Pro ($20/month): 160,000 monthly credits for daily use and building custom recipes.

Pros and Cons

Pros:

• Tested methods: each recipe lists its tested inputs, expected results, and known limits.
• Wide support: runs inside ChatGPT, Claude, Cursor, Codex, Copilot Studio, and other apps.
• Direct use: you can browse the marketplace and run any recipe yourself.

Cons:

• Credit math: costs vary by task, which makes monthly spending hard to predict.
• Small free tier: the free plan covers only about five small runs monthly.
• Simulated answers: survey simulations are hypotheses, not a replacement for real research.

FAQs:

1) What is a recipe?

An executable AI workflow with fixed inputs, steps, model choices, checks, and outputs.

2) Do I need an AI agent?

No, you can browse the marketplace and run recipes in your browser.

3) Can my agent build recipes?

Yes, it agrees inputs and outputs with you, then builds and tests a private recipe.

4) How does a recipe differ from a skill?

A skill gives an agent instructions, while a recipe runs fixed steps and checks.

5) Is there an API?

Yes, a Python SDK and API create runs and return structured JSON results.