AI Systems & Production Engineering
RAG pipelines, workflow automation, and SaaS architecture — practical notes for founders who need systems that ship, not slide decks.
Featured
Inkling by Thinking Machines Lab: Inside the New 975B-Parameter Open-Weights Multimodal Model
Thinking Machines Lab has released Inkling, a 975B-parameter Mixture-of-Experts model with 41B active parameters, native multimodality, and controllable reasoning effort — fully open-weighted and fine-tunable on Tinker. Here's a data-backed breakdown of its architecture, benchmarks, and what makes it different from other open-weights releases.
Haga: technical methodology and usage flow for independent physical-AI verification
How Haga evaluates robot policies and world models: simulation stress tests, physics-consistency checks, reproducible reporting, and intake flow.
Cursor Compile 26: Inside Cursor's First Conference and What It Means for AI-Native Development
Cursor held its first-ever conference, Compile 26, at Fort Mason in San Francisco. Here's a full recap of what was announced — Origin, Cursor Mobile, a new frontier model, and what it all signals for the future of software development.
Anthropic Just Found a 'Workspace' Inside Claude's Mind — Here's What It Means
Anthropic's new interpretability research uncovers a small internal region in Claude — the 'J-space' — that behaves strikingly like the brain's global workspace. Here's what it is, how it was found, and why it matters for AI safety and consciousness research.
Stop Prompting AI. Start Building Loops. Why Boris Cherny Thinks the Future Is AI Orchestrating AI
If you run a FinTech, HealthTech, or B2B SaaS company, the AI question is not "what is the best prompt?" — it is whether your systems can plan, execute, verify, and recover autonomously. Boris Cherny (Claude Code, Anthropic) builds loops, not one-shot prompts. Here is what that means for production AI.
Why independent verification is becoming the missing layer for physical AI
As physical AI moves from demos to deployed systems, verification is becoming the hidden dependency most teams still don’t have. This post explains why independent benchmarking matters now.