Program starts
Mon, Oct 5, 2026
Format
6 weeks · live + async
Time commitment
5–6 hrs / week
Taught live by
Howard Vance
ex-Staff PM, Anthropic Applied
0
0
Alumni
0.0
0.0
Rating
0%
0%
Completion
Alumni ship at
01 — The shift
Six weeks from now you are the person who settles the argument with numbers.
01
Scope a feature that survives contact
A spec with named failure modes, a fallback path and a kill criterion. Most AI features ship without all three and die of it by month two.
02
Build evals an engineer will actually run
Forty scored cases pulled from real transcripts, versioned in the repo. The numbers end the argument instead of starting it.
03
Price the feature before finance does
Unit economics end to end, three scenarios, caching and routing accounted for. You arrive at the budget review with the sheet already built.
04
Kill the wrong project in week one
Name the cases where a deterministic system beats a model, say it in the roadmap review, and bring the evidence that closes it.
02 — Your instructor
Howard Vance
ex-Staff PM, Anthropic Applied
Shipped applied-AI product to 40 million users. Teaches the part nobody documents: the judgment.
Howard led applied-AI product at Anthropic and ran the ML platform team at Figma before that. He has written the eval harness, defended the cost model to a CFO, and killed two features that demoed beautifully. This bootcamp is the internal onboarding he never got, compressed into six weeks and taught live, so the uncomfortable questions get asked while the decision is still open.
03 — Curriculum
Six weeks, six shipped deliverables
04 — Format
Five to six hours a week, always on the same three days
Weekly split
Live sessions
2.5 hrs
Project work
2 hrs
Reading & async review
1 hr
Every live session is recorded within two hours. Miss one and you keep the breakout notes plus a 20-minute async catch-up from Howard.
Weekly rhythm
MON
Live teaching session
09:00 PT
WED
Workshop & peer critique
09:00 PT
THU
Open office hours (optional)
16:00 PT
FRI
Deliverable due
17:00 PT
05 — Fit
Built for a specific kind of engineer
Enroll if you
Own a product surface where AI is already on the roadmap
Can hold five hours a week for six weeks, two of them live at 09:00 PT
Would rather be corrected by peers than agreed with politely
Have a live feature to bring as your project
Skip this if you
Want a certificate more than a shipped deliverable
Need an introduction to what a language model is
Plan to watch the recordings and skip the sessions
Are shopping for an implementation vendor, not your own judgment
06 — Results
What alumni shipped in the ninety days after
Killed a six-month AI roadmap item in week three
The eval work showed our retrieval was the problem, not the model. We shipped a boring fix instead and saved two engineers a quarter of work.
Priya Raman
Director of Product, Vela Health
Cut inference cost 61% in eight weeks
I walked into the budget review with the scenario sheet from week four. Finance approved the feature in one meeting instead of three.
Tobias Lund
Senior PM, Orbital
Got promoted to lead the AI surface
My roadmap defence from week six became the actual plan. I presented the same deck internally two weeks later and was given the team.
Alina Costa
Group PM, Figment
Shipped an assistant to 90k users with zero rollbacks
The launch runbook is the only reason. We caught two bad-output classes in staged rollout that would have been a public incident.
Marcus Oyelaran
Staff Engineer, Northlane
Turned a stalled pilot into a paid product in 70 days
We had been in pilot purgatory for a year. Scoping it properly with a kill criterion is what finally got it out the door.
Hana Ito
Head of Product, Kestrel AI
Rewrote how our design team specs AI flows
Failure modes are now a required section in every spec we write. That single change came out of week two.
Dev Shah
Design Lead, Ramp
07 — Pricing
Same bootcamp.
Two levels of access to Howard.
CORE PROGRAM
$2,400
one-time
or 3 × $840, no interest and no credit check
Six live sessions with Howard, capped at 24 people
Weekly workshop and a standing peer critique group
Six graded deliverables, ending in your roadmap defence
The eval harness, cost model and incident runbook templates
Permanent access to recordings and the alumni Slack
Full refund through the end of week two. No forms, no call.
6 Seats
PROGRAM + MENTORSHIP
$3,900
one-time
or 3 × $1,350, employer invoicing on request
Everything in Core Program
Three 1:1 sessions with Howard on your own live feature
Written review of your spec, eval set and roadmap
Direct Slack access to Howard for all six weeks
A session 90 days out to review what actually shipped
If the mentorship seats fill you move to Core and we refund the difference.
08 — FAQ
Before you apply
How much time does this really take each week?
Five to six hours: one 2–3 hour live session, roughly two hours of project work, and an hour of reading. Alumni who treated it as a two-hour-a-week course did not finish the final project. That is the honest number.
What happens if I miss a live session?
Recordings are posted within two hours with timestamps and the breakout notes. For the two workshop-heavy weeks, Howard records a 20-minute async catch-up covering what the bootcamp argued about, and you can bring your work to the next office hours.
Do I need to be able to code?
No, but you need to be comfortable reading a Python notebook and running a script someone else wrote. Product managers and designers do fine. If you have never touched a terminal, budget an extra hour in weeks three and four.
What is the refund policy?
Full refund through the end of week two, no forms and no exit call, just one email to hello@appliedai.co. After week two the material is substantially delivered, so refunds stop.
Can my employer pay for it?
Yes. We invoice with a PO, W-9 and itemised receipt, and about 70% of each bootcamp is employer-funded. There is a one-page justification memo in the pre-enrollment pack you can forward to your manager.
How is this different from a self-paced course?
Twenty-four people, one instructor, a fixed six weeks and a deadline every Friday. You are critiqued by peers doing the same job, and nothing is pre-recorded except the recordings of your own bootcamp.
The window closes Oct 2.
Applications close as seats fill, not just on the date.
Next program: 24 seats
Full refund through week two
