Slide contents

1. 1. AI Implementation in Grigory Oster Style

   Bad advice for those who want to spend the budget and change nothing

2. 2. 00. Why Grigory Oster?

3. 3. 'Bad Advice' as a Method

   Learning by contrast

   Oster: absurd advice for children

   The reader understands what NOT to do

   AI resembles old digital transformations

   Same failure patterns, new tech

4. 4. 01. Find a CAIO

5. 5. Chief AI Officer — The First Step

   Hire someone with a 'transformation' face

   Ops, sales, support optional

   What matters: roadmap through 2030

   Promises LLM, analytics, culture synergy

6. 6. 02. Allocate a Bigger Budget

7. 7. Less Understanding, Bigger Budget

   Small budget forces hard questions

   What process? What metric? Who's accountable?

   Big budget creates seriousness

   Consultants, licenses, workshops, stickers

8. 8. The Feeling of Progress

   Respectable corporations value motion vibes over motion

9. 9. 03. Create a Separate AI Organization

10. 10. AI Center of Excellence

    Isolation guarantees failure

    Separate room, chat and KPIs

    Business talks only through presentations

    Business: AI lives with specialists

    Specialists: business is not ready

11. 11. The Perfect Balance

    Nobody interferes with anyone's ability to change nothing

12. 12. 04. Move on to Promises

13. 13. Broad Statements Are Harder to Verify

    AI cuts routine 80%, speeds decisions 10x

    Democratizes expertise, personalizes CX

    Never specify: contract 2h → 20m

    Specifics kill corporate poetry

14. 14. 05. Throw a Festival

15. 15. AI Week and 142 Use Cases

    'Everyone is a prompt engineer'

    Bots for processes nobody has seen

    90 duplicate cases; 20 lack data

    More cards, more maturity

16. 16. 06. Don't Touch the Foundation

17. 17. Spread AI Like Mayo

    Work backstage

    Don't change processes

    Don't fix data or unify systems

    Don't clarify roles, KPIs or incentives

    Don't ask why

18. 18. AI Assistant Inside Chaos

    3 days → 5, with AI

    14 steps, 5 approvals

    AI drafts version one

    Then: manager, legal, brand, security

    Beautiful modern tech

19. 19. 07. Measure Activity, Not Impact

20. 20. Measure Activity, Not Impact

    'Maturity' Metrics ✓

    How many licenses purchased

    How many pilots launched

    How many employees trained

    Impact Metrics ✗

    Process cycle time

    Error rate

    Cost per operation

21. 21. AI Strategy Disconnected from Bonuses

    Innovation vs self-preservation

    Bonuses reward manual control.

    From above, talk automation and delegation.

    In presentations, everyone loves AI.

    In practice, status tasks stay hidden.

22. 22. 08. One Year: Still a Phase

23. 23. The Reboot Ritual

    Early adoption stage

    Culture and data need more time

    Restructure the AI office

    Another manifesto, budget increase

    Buy an enterprise-agent platform

24. 24. 09. The Uncomfortable Truth

25. 25. AI Is Not Frosting

    If decisions stay slow

    If data is still scattered

    If accountability is still blurred

    If the process still passes papers around

    AI speeds drafts, not reform

26. 26. AI Didn't Fail — the Attempt Did

    AI on old order

    How people work and make decisions

    How they share responsibility

    This technology can't be purchased

    It's called reform

27. 27. What to Read to Avoid the Trap

    'From Classic PDLC to AI-Native Development'

    'From AI-Native Development to AI-Native Organization'

    Big Tech: Amazon, Meta, Microsoft, Google

28. 28. References and Materials

    For adopting AI the right way

    Grigory Oster — 'Bad Advice'

    'From Classic PDLC to AI-Native Development'

    'From AI-Native Development to AI-Native Organization'

29. 29. Thank You!

    Bad Advice on AI

    For more materials on this topic, visit the "Book Cube" channel — all links are collected there

    Alexander Polomodov, Technical Director & Fellow, T-Technologies

    @Book_Cube