What it takes to build production ready AI solution
1 like425 views
The document discusses the challenges of building a production-ready AI solution, emphasizing the importance of setting clear expectations and processes. It shares a story of an SEO marketer facing unrealistic demands, drawing parallels to the common misconceptions around AI capabilities. Key points include the significance of defining use cases and success criteria, along with the need for proper data to ensure project success.
What it takes to build production ready AI solution
- 1. WHAT IT TAKES TO BUILD A PRODUCTION READY AI SOLUTION
- 2. HELLO
- 4. DEALING WITH NEW TECH: 1. EXPECTATIONS 2. PROCESS
- 6. HARD WORKING BUT FRUSTRATED SEO MARKETER
- 7. THE MOST HATED QUESTION IN SEO:
- 8. “Bring me to the top of the Google search page.”
- 9. SOME THINGS THAT ARE EASIER TO DO
- 14. IT’S A SIMILAR STORY WITH AI
- 15. HOW TO AVOID .HATED. REQUESTS?
- 17. HOW DO YOU SET EXPECTATIONS?
- 18. ANDREW NG AI FOR EVERYONE
- 19. NEW TECH HOW TO WORK WITH
- 21. “I want to schedule my meetings in the app!”
- 24. BUT WHAT IF YOU’VE GOT SOMETHING SO NEW YOU DON’T KNOW WHAT TO DO?
- 25. NO EXPECTATION NO PROCESS ...NO PROBLEM?
- 28. WE LEARNED THIS THE HARD WAY
- 29. OUR PROCESS DISCOVERY WORKSHOP IDEATION PHASE TECH DUE DILIGENCE PROOF OF CONCEPT DATA SOLUTION
- 30. THE PITFALLS
- 31. USE CASE IS NOT CLEARLY DEFINED
- 32. SUCCESS CRITERIA IS NOT CLEARLY DEFINED
- 34. LACK OF DATA
- 35. NON-LABELED DATA
- 37. STALE DATA
- 39. TUNING PHASE
- 40. RUNNING COSTS
- 41. HUMAN IN THE LOOP
- 42. POST-DELIVERY
- 43. DATA SOLUTIONS ARE FOR PEOPLE
- 44. SUCCESS RATE
- 45. OR CONCLUDE THAT AI IS NOT FOR YOU. YET.
- 46. THANK YOU