Collinear AI’s Blog
Subscribe
Sign in
Home
Archive
About
Introducing Collinear Simulations: Steerable Personas for AI Agent Testing
TraitBasis, inspired from mech intrep, gives high-fidelity user personas for comprehensive agent testing
Oct 7, 2025
•
Meghana A Rajeev
,
Tsach Mackey
,
Muyu
,
Anand Kumar
, and
Nazneen Rajani
2
Latest
Top
Discussions
SimLab: The self-serve staging playground for real-world agents
Agents fail on real tool calls, long workflows, and messy data. SimLab lets you find those failures in simulation, not in production.
20 hrs ago
•
Sachin
9
Collinear Newsletter #10 - Notes on Frontier AI
Happy March from the Collinear team!
Mar 24
•
Soumyadeep Bakshi
and
Nazneen Rajani
10
We gave Claude, Gemini and GPT, $250k, and it didn't go as you’d expect...
Introducing YC Bench: The first open-source, long-horizon benchmark with a simulation clock
Mar 5
•
Muyu
and
Anand Kumar
6
Collinear Newsletter #9 – Notes on Frontier AI
Hi AI innovators,
Dec 19, 2025
•
Soumyadeep Bakshi
and
Nazneen Rajani
5
RL Infrastructure for AI Agents: Why Environment-as-a-Service is the Missing Piece
Reinforcement learning for large language models is more of a systems problem than ML.
Nov 18, 2025
•
Nazneen Rajani
3
1
Announcing Spider: a lightweight tool to craft post-training data recipes
TL;DR Spider is a single client interface that turns messy distillation and ablation experiments into a simple, configurable workflow.
Nov 6, 2025
•
Soumyadeep Bakshi
6
Collinear Newsletter #8 – Notes on Improving AI
Hi AI innovators,
Nov 4, 2025
•
Soumyadeep Bakshi
8
See all
Collinear AI’s Blog
Building the Enterprise AI Improvement Data Flywheel
Subscribe
Collinear AI’s Blog
Subscribe
About
Archive
Sitemap
This site requires JavaScript to run correctly. Please
turn on JavaScript
or unblock scripts
