▶ 79:00 · Developer Tools · Auto-research agents running nightly optimization experiments across entire codebase

episode briefing

AI debate with Factorial engineers | itnig discussion

2026-04-03 · 6 company · 6 thematic

sentiment

3 bull2 bear1 neu

speakers

Staff engineer at Factorial who hasn't written code directly in months, focusing on system-level thinking and architecture. Built an orchestrator to understand agent decision-making. Views himself as a bottleneck for PR reviews.

miguel

Engineering lead at Factorial driving the company's AI-native transformation. Led the cultural shift to zero-code experiments, standardization of AI tooling, and adoption of auto-research agents. Strong views on Software 2.0 transition and model provider economics.

adria

Factorial engineer building custom dashboards using Jira as a data backend only. Experiments with Talis hardware and auto-research agents. Skeptical of pure prompt-based software due to cost, favors hybrid deterministic + LLM approach.

episode shorts · 4

Reviewing vs producing code

Can a child be a programmer?

Software 2.0

The programmer's flow

now playing · Developer Tools

AI Economics & Business Modelstailwindscore 8/10miguel

Model providers subsidizing inference 10-13x to own the application layer

Anthropic (Claude Code) and likely OpenAI are pricing consumer coding products far below API costs (13x subsidy) to drive adoption and lock in developers. This vertically integrated model t…

AI Coding Agentstailwindscore 8/10miguel

Code review becoming the bottleneck as AI generates 10x more PRs

AI coding agents have shifted the bottleneck from code generation to code review. Factorial sees 150+ PRs pending review (up from 30) because human review doesn't scale. They are investing…

AI Infrastructuretailwindscore 9/10miguel

Software 2.0 transition: codebases becoming prompts + context + skills

The industry is transitioning from Software 1.0 (deterministic code) to Software 2.0 (prompts, skills, context files executed by LLMs). Companies will maintain prompt/skill repositories ins…

Open Source AImixedscore 7/10adria

Open source shifting from code distribution to prompt/skill sharing for LLM training

Open source value is evolving: using standard frameworks (React) gives advantage because LLMs are trained on them. Future open source will be shared prompts, skills, and context files that…

Developer Toolstailwindscore 7/10miguel

Auto-research agents running nightly optimization experiments across entire codebase

Factorial is deploying 'Auto Research' (inspired by Karpathy's repo) where agents run hundreds of parallel experiments overnight (performance, structure validation, etc.) and submit PRs wit…

Enterprise AI Adoptiontailwindscore 8/10edu

AI-native product development: generating UIs on-demand per user request

Factorial's 'One' agent generates custom dashboards and analytics UIs in real-time from natural language queries, moving from pre-built deterministic screens to generative UI. This reduces…

←