/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

Q&A with Simon Willison on the November release of GPT-5.1 and Opus 4.5 as the inflection point for coding, exhaustion due to managing coding agents, and more

Simon Willison is a prolific independent software developer, a blogger, and one of the most visible and trusted voices on the impact AI is having on builders.

Lenny's Newsletter Lenny Rachitsky

Context & Ripple Effects

Coverage moved from mixed evidence on GPT-4.5 coding performance to reports that GPT-5 improved at practical software-engineering work, while an earlier hands-on assessment found GPT-5 competent rather than categorically ahead. Willison’s view that the November models changed coding therefore matters less as a single benchmark claim than as a practitioner signal that capability may have crossed a workflow threshold.

The same account highlights the cost of that shift: managing coding agents can itself become exhausting. That tension is consistent with later emphasis on long-horizon code changes and context management—areas where greater agent autonomy also creates more work to direct and verify.

First-order effects

  • Developers who treat GPT-5.1 and Opus 4.5 as capable coding agents can delegate larger implementation tasks, but must spend more attention defining work, reviewing output, and recovering from agent mistakes.
  • For independent builders such as Willison’s audience, agent management becomes a practical throughput constraint alongside model quality; adoption is not simply a matter of prompting once and accepting the result.

Second-order effects

  • Model providers compete not only on coding capability but on reducing coordination overhead: durable context, task handoffs, observability, and review workflows become more consequential product surfaces.
  • Software teams are pushed toward clearer task decomposition and stronger validation practices, because the reported coding gains make delegated work more attractive while the reported fatigue raises the cost of poorly governed delegation.

Third-order effects

  • If coding agents continue to take on longer tasks, software development may shift from individual code production toward supervising a portfolio of agent-driven changes; the winning workflow will be the one that makes that supervision reliable rather than merely fast.
  • The pattern strengthens the case for operational assurance around agentic development: organizations will need to make accountability, testing, and human sign-off scale with the volume of AI-generated changes.

The trend: Coding AI is moving from autocomplete-style assistance toward agentic implementation, with coordination and assurance becoming the limiting factors for real productivity gains.

Discussion

  • @lennysan Lenny Rachitsky on x
    “Using coding agents well is taking every inch of my 25 years of experience as a software engineer, and it is mentally exhausting. I can fire up four agents in parallel and have them work on four different problems, and by 11am I am wiped out for the day. There is a limit on [vid…
  • @levelsio @levelsio on x
    If he'd take 10g creatine he could extend it to at least 2pm
  • @josevalim José Valim on x
    The best analogy of how I use coding agents is pair programming. They write the code, I review it. They write the spec, I think about the next problem (sometimes using agents too!). Etc. The only time I ever run more agents (3+) is when I absolutely don't care about the code
  • @addyosmani Addy Osmani on x
    Tip: Figure out your personal ceiling for running multiple agents in parallel. We need to accept that more agents running doesn't mean more of _you_ available. The narrative is still mostly about throughput and parallelism, but almost nobody's talking about what it actually
  • @nbaschez Nathan Baschez on x
    This is very true in my experience Programming used to feel kind of leisurely and relaxing, like a crossword or sudoku Now it feels more like debate. My raw horsepower is at the limit of my ability to absorb complex information and make decisions with real consequences.
  • @simonw Simon Willison on x
    I was a guest on @lennysan's podcast! We talked about agentic engineering and all sorts of other LLM-related topics for 1h39m(!), plus a little bit about kākāpō parrots - here's my selection of highlights from our conversation https://simonwillison.net/...
  • @provisionalidea James Rosen-Birch on x
    it is great to see some sanity around the fact that agents increase the need for solid computer science fundamentals and software engineering skill, and using them well takes effort as opposed to the frenetic and hypoxic claims the machines will replace all coders
  • @hugobowne Hugo Bowne-Anderson on x
    simon's spot on here and i'm glad someone so prominent is stating this publicly. on top of this, coding agents and the human reward system interact in novel and strange ways we'll be unpacking for years to come imo so stay safe out there, people!
  • @paulabartabajo_ Pau Labarta Bajo on x
    I can relate so so much to this. I would also stress, let's not forget about “enjoying” the process of building, and not just the destination. These tools bring us so much closer to the destination, that we might fall into the trap of forgetting about the journey.
  • @luizajarovsky Luiza Jarovsky, PhD on x
    🚨 Most of the AI industry seems to ignore what @simonw is saying here. There is a biological LIMIT to humans' cognitive load, and pushing beyond it will likely lead to lower-quality decision-making and ethical/legal risks. I call it the acceleration paradox (more info below).
  • @lennysan Lenny Rachitsky on x
    My biggest takeaways from @simonw: 1. November 2025 was an inflection point for AI coding. GPT 5.1 and Claude Opus 4.5 crossed a threshold where coding agents went from “mostly works” to “almost always does what you want it to do.” Software engineers who tinkered over the
  • @lennysan Lenny Rachitsky on x
    I asked @simonw what the next leap in AI software engineering is likely to be. He explained the “dark factory” pattern where teams don't write any code or even look at their code. [video]
  • @lennysan Lenny Rachitsky on x
    “Using coding agents well is taking every inch of my 25 years of experience as a software engineer.” Simon Willison (@simonw) is one of the most prolific independent software engineers and most trusted voices on how AI is changing the craft of building software. He co-created
  • @pmarca Marc Andreessen on x
    AI increases workload. Many such cases.