Building with AI: Flo Crivello on Lindy.ai
Links and Chapters below:
Flo Crivello on X (Twitter) - https://x.com/Altimor
Lindy.ai - https://www.lindy.ai/
00:00 - Introduction
02:44 - prompting poor mans RLHF
05:30 - AI self critique
08:20 - mental model for AI
09:00 - can’t self correct
10:45 - using LLMs as classifiers
12:55 - when will users trust AI
14:32 - power of k-shot examples
16:25 - buildtime vs runtime
18:27 - conditional branching and cost reduction
20:32 - “gpt wrappers”
21:50 - building Lindy.ai
22:43 - evals
24:28 - temperature zero
26:26 - open weights LLMs
27:51 - fine-tuning
28:36 - 6 credible LLM players
30:35 - capability overhang
31:07 - building with AI, easier or harder?
33:08 - just ask the model
33:35 - Lindy.ai team
34:29 - inference optimization
37:06 - reliability engineering
41:28 - a world where you can talk to your computer
41:19 - general purpose data processing machines
43:58 - AGI, now what?
45:59 - the early days