Sayantan Das
Subscribe
Sign in
Humans Aren't Leaving the Loop
Agents compress the execution cycle. Humans move up to the strategic one.
Jun 4
•
Sayantan Das
Log-Centric Agent Architecture
How event logs, deterministic projections, and replayable state change the way advanced agents are built
Jun 4
•
Sayantan Das
May 2026
Evals Are a Feedback Loop, Not a Score
What the best agent teams in 2026 know about measuring their systems
May 20
•
Sayantan Das
How I Run Long-Running Agents at an Insurance Company
Durable orchestration, checkpointing, and human-in-the-loop in production
May 20
•
Sayantan Das
1
They Said Extensions Were Dead. Then AI Needed a Browser.
The arc of browser plug-ins — from power-user toy, to security liability, to the quiet backbone of the agentic web.
May 18
•
Sayantan Das
Agent Design Patterns Used in Enterprise AI Systems
What I've learned building LLM agents that actually run in production
May 17
•
Sayantan Das
Be an Internal FDE for Your Company
The most underrated career move in data science isn't a promotion. It's becoming the person everyone already trusts.
May 17
•
Sayantan Das
How to Build a Killer LLM Studio in 2026
A system design deep-dive: six services, one gateway, and the engineering decisions that actually matter
May 17
•
Sayantan Das
11 tools I built for myself
Small utilities that do one thing. Zero onboarding.
May 17
•
Sayantan Das
I made my Claude agent unkillable on macOS
Three layers of self-healing for a Telegram bot that runs forever
May 17
•
Sayantan Das
The legal system wasn't built for you. Sulajh is.
AI-powered dispute resolution for the disputes that never make it to court.
May 17
•
Sayantan Das
This site requires JavaScript to run correctly. Please
turn on JavaScript
or unblock scripts
