🌟 Access the Dashboard
🎣 What's New: Fresh weekly catch
We refreshed our sources for the November 8 edition—including Dear Architects #315, Token by Token #30, Andriy Burkov AI #350, Editor's Radar, and our regular engineering blogs—and pulled together grounded reads on agentic loops, weight-space programming, distributed systems, and engineering leadership:
- The Missing Link for Speed with Quality — Adam Tornhill (Token by Token #30 · Spotlight) & Beyond the Harness: Canvas for AI Agents — Nimisha Asthagiri (Dear Architects #315 · Token by Token #30)
Why early AI productivity gains vanish when agents hit legacy technical debt, paired with Nimisha Asthagiri on designing a bounded architectural canvas with drift detection and decision records so AI agents can operate with safe autonomy.
- Practical Loop Engineering — Addy Osmani (Editor's Radar) & Five months treating bugs like patients and coding agents like a medical team — Adam Storm & Rafi Shamim (Burkov AI #350)
Two hands-on playbooks for autonomous loops: Addy Osmani on shifting from turn-by-turn prompting to self-correcting verification loops, and Cockroach Labs' "MOLT Sinai" case study on structuring coding agents like a teaching hospital across 1,000+ pull requests.
- How Meta Built Forum With AI-Native Engineering — Meta (Dear Architects #315) & How software engineering is changing: a tale of two eras — Neil Kakkar & Gergely Orosz (The Pragmatic Engineer)
How a small Meta team built a standalone Groups app to feature parity in eight weeks by treating AI agents as supervised operators, alongside Neil Kakkar's contrast between pre-AI shipping (8–20 PRs/month) and AI-native teams merging 200+ PRs/month.
- Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights — Rikiya Takehi, Jayden Teoh, Alex Zhang & Omar Khattab (Editor's Radar · Burkov AI #350) & Qwen 3.8 27B on my laptop can interpret model weights, too — Simon Willison (Simon Willison's Weblog)
MIT CSAIL shows that frontier LLMs can directly inspect and edit raw neural network weights—fixing a reasoning flaw in Qwen3.5-0.8B by editing just 0.016% of its parameters—followed by Simon Willison replicating the weight-reading experiment locally on his laptop.
- Why Antithesis? — Hillel Wayne (Editor's Radar) & Evals for AI Engineers — Shreya Shankar & Hamel Husain (Token by Token #30)
Hillel Wayne on how Deterministic Simulation Testing (DST) bridges formal methods and real-world codebases, paired with Shreya Shankar and Hamel Husain's practical handbook on building systematic evaluation pipelines for production LLM systems.
- Educating an Architect: A Conversation with Luca Mezzalira — Luca Mezzalira & Michael Stiefel (Dear Architects #315 · Spotlight) & The Data Product Platform Navigator — Tom De Wolf (Martin Fowler)
Software architecture as a socio-technical discipline balancing system design with organizational topology, alongside Tom De Wolf's four-level maturity navigator for Data Mesh and self-service data product platforms.
- Scaling Slack and GitHub with Maude Lemaire — Gergely Orosz (The Pragmatic Engineer) & Building a large-scale distributed storage system based on Raft — Edward Huang (Dear Architects #315)
Maude Lemaire on enterprise tenant sharding, $150,000 load tests at Slack, and scaling GitHub's Rails monolith, paired with PingCAP CTO Edward Huang on multi-Raft distributed storage architecture.
- The Multiplier Mindset: How to Move from Senior Engineer to Tech Lead in the Age of AI — Gregor Ojstersek (Editor's Radar) & From writing code to leading through others — Tomasz Borys (Token by Token #30)
Two thoughtful perspectives on stepping into technical leadership and multiplying team clarity, ownership, and velocity when writing code yourself is no longer the primary bottleneck.