Resources

Free guides and reports for teams adding AI to the software they already run.

Each guide and report answers one question we hear from product, engineering and IT leaders, with the checks and worksheets we use in our own work. Read the first pages free, then send a short form for the rest and the PDF.

Start here

Where AI most often fails once real customers use it, how to find it first, and how to answer the security review that follows.

Decide

Where AI should start in your product, and what to check first.

  • Report · 14 pages

    We tagged 52 AI projects. Here is what they needed.

    State of AI features: what 52 projects needed

    This answers

    • What did real AI features need, beyond the model?
    • Where did the time go, and what got decided late?
    • What should that change on our own AI roadmap?
    Get the report
  • Worksheet · 8 pages

    The top idea in our worked example needed no AI.

    Pick your first AI feature: code, search, model or agent?

    This answers

    • Which steps in our product need AI at all?
    • How do we rank a long list of AI ideas?
    • Which feature should we build first?
  • Checklist · 8 pages

    Your API docs describe the system you planned.

    Before you build AI: inspect the system, agree on done

    This answers

    • What must we know about our own product first?
    • What does done mean, in lines anyone can test?
    • Who decides what, before anyone writes code?
    Get the guide
  • Template · 8 pages

    Your board wants one page on AI, not a deck.

    The one-page AI board brief

    This answers

    • What do I tell the board about our AI plan?
    • What ships first, and when will customers have it?
    • How do we show progress and risk each quarter?
    Get the guide
  • Scorecard · 8 pages

    Most products should start hosted. Score yours before buying GPUs.

    Self-host a model or call an API? Score it in 10 lines

    This answers

    • Can our customers' data stay in our own cloud?
    • Do we need a private model, or is an API enough?
    • What does running our own model really take?
    Get the guide

Build

How to connect AI to your product safely: your data, your permissions, your older code and the actions it takes.

  • Blueprint · 8 pages

    Your agent retried a payment. The customer got paid twice.

    Before your AI agent touches money: limits, approvals, undo

    This answers

    • Which actions may an agent take alone, and who approves the rest?
    • What happens when a payment times out or a step fails halfway?
    • How do we stop text the agent reads from steering what it does?
    Get the guide
  • Blueprint · 8 pages

    AI adds five new places where two customers' data can meet.

    Can customer A ever see customer B's data in your AI?

    This answers

    • Where can customer data leak once AI is in the product?
    • How should search keep each customer's results apart?
    • What should fail closed, and how do we set it up?
    Get the guide
  • Checklist · 8 pages

    Your users will ask AI about your product either way.

    Will your product work inside ChatGPT and Claude?

    This answers

    • What would our users see in ChatGPT or Claude?
    • Are we ready, and which three tools come first?
    • How does it stay under our login and permissions?
  • Report · 15 pages

    A signed BAA is the start, not the finish.

    HIPAA-ready AI features for healthcare software

    This answers

    • What does an AI provider's BAA cover, and what does it leave to us?
    • How do we keep patient data out of prompts and logs?
    • What will clinics ask about the AI before they switch it on?
    Get the report
  • Blueprint · 9 pages

    Never give an agent run_sql. Give it find_past_orders.

    Side panel or thin task layer: AI in the app you already run

    This answers

    • Where should AI sit in our existing screens?
    • What if our app is all pages and stored procedures, with no API?
    • What does the agent get instead of raw access to our data?
    Get the guide
  • Explainer · 8 pages

    Your dealers want AI. Their data decides what works.

    AI on top of the DMS: 10 dealer workflows and their data

    This answers

    • Which dealer workflows suit AI first?
    • What data does each one need from the DMS?
    • How do we sell it to dealers and keep each dealer apart?
    Get the guide

Prove

How to test it, launch it and keep it working.

  • Report · 15 pages

    One prompt fix can quietly break answers you never re-checked.

    Stop testing AI by vibes: the eval plan template

    This answers

    • How do we test an AI feature before it ships?
    • What goes in a golden set, and what blocks a release?
    • How do model upgrades stay safe?
    Get the report
  • Checklist · 8 pages

    Your pilot works. So why is it still a pilot?

    From pilot to product: the AI go-live sheet

    This answers

    • Why do AI pilots stall before they ship?
    • What must be true before customers rely on our AI?
    • How do we roll it out and watch the first 30 days?
    Get the guide

Sell and scale

How to price your AI, answer your customers' questions about it, from their security review on, and get it used once it is live.

  • Report · 15 pages

    Your AI won the pilot. The privacy agreement decides the deal.

    The district AI procurement pack

    This answers

    • Can our AI touch student records, and on what terms?
    • What will districts ask before they sign?
    • What goes in the data privacy agreement and the parent note?
    Get the report
  • Report · 15 pages

    Your AI feature is free to customers. Is it free to you?

    Will this AI feature pay?

    This answers

    • Is it worth building?
    • What will it cost to run, per customer?
    • How do we price it without losing margin?
    Get the report
  • Checklist · 8 pages

    The feature works. It sits where nobody looks.

    We shipped AI and nobody uses it

    This answers

    • Why do customers ignore our AI feature?
    • Where should it sit in the product?
    • How do we measure adoption and fix it?
    Get the guide

Your customers already expect AI from your product.

Let us make it real inside the product you run today.