AI Boost X
A curated LLM eval package with a benchmark
A curated eval set and an open benchmark: find the model, effort and cache setting that clears your bar at the lowest cost.
Tag: ai-tools
Riding two AI horses: Claude Code and Rovo
Claude Code sees the code. Rovo sees the org. With me as the harness, this approach got a design to near perfect.
The harness is the team: one day at Google Cloud's Agents for Impact Hackathon
Everyone at the table had AI agents running. I launched one new session, and it joined a team of agents already set up with roles. Plug and play, and portable. We built RescueBroker in one day.
Why 100x is easier than 1x
How 100x happens, case by case, and a playbook to try on Monday.
Sonnet 5.5 another game changer, Benchmarked with Gemini 3.8 Flash and Argon
Sonnet 5.5 outperformed Opus 5.5 and Fable 5.1 on speed and cost
How I build the Agent Harness, 10,700 lines changed in hours while keeping the quality bar
Four AI agent sessions built most of a production feature in one working day: 85 files, 46 commits, 614 passing test runs, merged the same afternoon after the client team's review. Here is how it was checked.
I built an Agent Framework Titan-Kronos: Implemented in Production at scale
On day one with Fable 5, I had it adversarially review an agent skill Opus 4.8 built. It found and fixed 12 real bugs, and two days later the skill was production-ready. The bill was $690.
Google Antigravity IDE for vibe coding, part 1
A LinkedIn post introducing Google's Antigravity IDE for vibe coding.
Google Antigravity IDE for vibe coding, part 2
A follow-up LinkedIn post on Google's Antigravity IDE for vibe coding.
Vibe coding faster and cheaper
A LinkedIn post on vibe coding faster and cheaper.