GuideAdvanced
Deciding Where to Use AI in Your Product Guide
Almost every software product gets the same order today: "add AI." The hard decision is not which model to use, but where AI adds value and where it doesn't, how much freedom you give it, how good it has to be before a customer sees it, who signs off on that bar, and when to turn it off. This guide teaches you to make those decisions from the leadership table: whether an idea needs AI or a rule is enough, what level of autonomy it acts with, what bet you sign before building, what experience it needs so users trust it, what quality bar it must meet, how to learn from its mistakes with real users, how much cost and how much risk the product accepts, and when to continue, pivot, or kill. You learn it as Head of Product at **Tallyfern**, an invoicing app for freelancers in Latin America and the United States. The CEO asks in the leadership meeting where to put AI, and sales is already promising a "collections autopilot." Module by module you decide which ideas get AI, sign the bet for **Nudge** (a payment reminder drafted in the freelancer's tone), approve its experience and its quality bar, read why the first beta collected less than the old template, decide what changes, set its cost and risk budget, and close with a continue, pivot, or kill memo that you defend in front of the CEO and the board. There is no code to write. The math is done in tables and by hand; technical reports (a quality bar, review notes, a dashboard) are shown so you can read them and decide, and the studies come with their source and date.
- lessons
- 64
- modules
- 8
- available in
- English · Spanish
- access
- Free

What you'll be able to do
- Decide whether an idea needs AI, classic machine learning, a rule, or should not ship, based on the cost of an error rather than on what the model can do
- Set the autonomy level of an AI feature and know how to say no to an "autopilot"
- Write and sign a feature's bet before building it, with model, experience, and business metrics, a baseline, and decision rules
- Treat the experience as a trust decision: disclosure, attribution of facts, editing, undo, fallback, and waiting
- Set a quality bar with gates, targets, and severity, and decide who signs each line
- Ask for and read an error analysis with real users and decide what gets fixed first and where
- Set how much cost and how much risk the product accepts for the feature to be worth it
- Deliver a continue, pivot, or kill memo and defend it in front of leadership
What you need to bring
It's for you if...
- Heads of product and directors or VPs of product who decide which features get AI
- Startup CTOs, technical founders, and heads of engineering who answer for the quality of the product's AI
- Tech leads and engineering managers who will take an AI feature from idea to decision
Requirements and materials
- Having made or supported product decisions
- Having used a generative AI tool; you do not need to code or know how to call a model
The syllabus, module by module
Open any of them to see its lessons.
- 1. Introduction: the board asks what we are doing with AI
- 2. A probabilistic feature in a product that lives on being exact
- 3. Who pays for an error
- 4. Every use costs money: variable cost as a constraint
- 5. The simplest thing that could work
- 6. The AI fit matrix
- 7. Autonomy levels
- 8. Project: AI fit memo for the CEO
- 1. Introduction: evidence before and after users
- 2. Evidence before you have users
- 3. Using it in-house and opening the first beta
- 4. From failure codes to checks
- 5. User behavior as evidence
- 6. Watching quality in production
- 7. Quality as a product metric
- 8. Project: the v1 versus v2 comparison and the evidence plan
- 1. Introduction: what the product accepts to pay and to risk
- 2. The cost of a call decides nothing
- 3. Which product decisions multiply cost
- 4. How much risk the feature accepts: the error budget
- 5. Usage limits and the users who consume the most
- 6. Guardrails when cost or risk go out of bounds
- 7. Value against cost and risk: is it worth it?
- 8. Project: Nudge's cost and risk budget
What people usually ask
No limit. It's a free guide: come in whenever you like, as often as you like.
No. Modules run from easier to harder, but you can jump to the one you need. Progress is saved per lesson.
Whatever is needed is listed under “What you need to bring”, above. If nothing is listed there, you can start from zero.
In the WhatsApp group, and every two weeks there's a live with an instructor where questions get worked through.
Start whenever you like
What students say
These reviews are from enrolled students who completed at least 50% of the course. We moderate reviews only on content grounds (spam, offensive language, personal data), never for being critical or negative.
No approved reviews yet.
Be the first to share your experience!