How to build AI products and ad products for what comes after the click.
By Nicolas DavidVP of Product
Episodes
- EP 03LiveSponsored answers
How to build sponsored answers in an AI assistant
Ad products · 1 min
- EP 04Next upAn eval system
How to build an eval system for an AI product
AI products
- EP 05WritingAttribution, no click
How to build attribution without the click
Ad products
Two tracks
Pick your track
AI products
3 episodesAgents, answer engines and copilots that people pay for. How to scope, evaluate, staff and price them.
- Agents
- Evals
- RAG
- Pricing
- Trust
Ad products
3 episodesAds when AI answers and agents act. Targeting, auctions and measurement for the post-click world.
- Contextual
- Auctions
- Sponsored answers
- Attribution
- Retail media
How I build
Four rules I use on every product
Pick the buyer first
Every product starts with who pays and what number they report on. Features come after.
Write the eval before the prompt
If you can't check the output, you can't ship it. Define good, then build.
Ship small, price early
A price on day one tells you more than a survey. Charge for the first version.
Own the go-to-market
A product nobody can sell is a demo. The VP of Product owns the pitch too.
Numbers
- 15
- Years in adtech
- 6
- Years as VP Product
- 3
- Episodes
About
I'm Nicolas David, VP of Product. I have spent my career building ad products, and now I build AI products too.
This site is where I write down how I do it, one playbook at a time.
More about me →