GenAI PMSimon Willison
Developer and writer known for his AI tooling commentary and the `llm` project. He is credited here with the 0.32a2 release note.
Key Highlights
- Simon Willison is a key interpreter of AI tooling changes, especially where model APIs affect developer and product workflows.
- He is credited here with the llm 0.32a2 release note, including support for OpenAI's /v1/responses endpoint and reasoning-token controls.
- His writing gives AI PMs practical guidance on migrations, output formats, coding agents, and cross-provider model tooling.
- He frequently connects technical releases to UX and workflow implications, making his work useful for roadmap and implementation decisions.
- His experiments with HTML output, browser-based parsing, and model comparisons provide concrete examples of applied AI product thinking.
Simon Willison
Overview
Simon Willison is a developer, open source builder, and prolific writer who is widely followed for clear, practical commentary on AI tooling, developer workflows, model capabilities, and product implications. In this corpus he is most directly credited with the llm 0.32a2 release note, and he appears repeatedly as a source for hands-on experiments, release coverage, and observations about coding agents, model interfaces, and real-world usage patterns.For AI Product Managers, Simon matters because he consistently translates fast-moving technical changes into concrete product lessons. His work often sits at the intersection of model APIs, UX, open source tooling, and developer adoption. That makes him a useful signal for PMs tracking how new model features actually change workflows, what implementation details matter in practice, and where agentic products create value or risk.
Key Developments
- 2026-04-24 — Simon adapted LlamaIndex's LiteParse into a browser-based PDF text extraction workflow, showing how document-processing tools can be pushed fully client-side using web-compatible libraries.
- 2026-04-24 — He was also cited among commentators covering the launch of GPT-5.5, reinforcing his role as a trusted interpreter of major model releases.
- 2026-04-25 — Simon was listed among those covering API access for GPT-5.5 and GPT-5.5 Pro, highlighting his relevance to builders integrating frontier models into products.
- 2026-04-26 — In his notes on the GPT-5.5 prompting guide, Simon emphasized migration tips, a Codex migration command, and the warning not to treat GPT-5.5 as a drop-in replacement for prior models.
- 2026-05-05 — Simon published a Granite 4.1 3B SVG Pelican Gallery, comparing outputs from quantized model variants and concluding that output quality did not show a simple relationship to model size.
- 2026-05-07 — He live-blogged Anthropic’s Code w/ Claude 2026 event, capturing keynote highlights and observations relevant to AI coding workflows.
- 2026-05-07 — He also reflected that vibe coding and agentic engineering are converging in uncomfortable ways, signaling concern about product and workflow tradeoffs in AI-assisted software development.
- 2026-05-08 — Simon announced llm-gemini 0.31, noting that gemini-3.1-flash-lite was no longer in preview, a practical update for teams relying on model stability and package support.
- 2026-05-09 — In Using Claude Code: The Unreasonable Effectiveness of HTML, Simon explored asking models for HTML instead of Markdown, highlighting richer outputs such as SVG diagrams and interactive widgets.
- 2026-05-12 — Simon highlighted Tobias Lütke’s description of Shopify’s public coding agent River, connecting visible workflows to faster organizational learning and comparing the pattern to Midjourney’s public Discord dynamics.
- 2026-05-13 — Simon was credited here with the llm 0.32a2 release note. The release added support for OpenAI models using the /v1/responses endpoint, enabling reasoning-capable models to interleave reasoning and tool calls. It also surfaced summarized reasoning tokens separately and added flags to hide reasoning when desired.
Relevance to AI PMs
1. He helps PMs understand what new model capabilities mean in product terms. Simon’s coverage of GPT-5.5, OpenAI’s responses API, Gemini releases, and Claude-related workflows is useful for translating model release notes into roadmap implications, migration plans, and UX decisions.2. He provides practical examples of developer-facing AI product design. Through projects like llm, llm-gemini, and experiments with HTML output, browser-based PDF extraction, and coding agents, he surfaces implementation details that PMs can use when defining tooling, extensibility, observability, and user controls.
3. He is a strong signal for workflow shifts in agentic software. His writing on public coding agents, vibe coding, and agentic engineering helps PMs think tactically about trust, transparency, handoff design, testing expectations, and when to expose or hide model reasoning in user experiences.
Related
- llm / llm-python-library — Simon’s best-known tooling connection in this dataset; a core example of open source infrastructure for working with multiple LLM providers.
- OpenAI / codex / gpt-5.5 / gpt-5.4 / responses API — Frequent connection point through Simon’s commentary on model launches, prompting guidance, and API behavior changes.
- Anthropic / Claude / Claude Code / Code w/ Claude 2026 — Connected through Simon’s event coverage and experimentation with coding workflows and output formats.
- Google / Gemini / llm-gemini / gemini-3.1-flash-lite / gemma — Relevant via plugin and release coverage, especially for teams tracking cross-provider support.
- LlamaIndex / LiteParse — Connected through Simon’s browser adaptation work, showing interest in practical document and parsing workflows.
- Tobias Lütke / Shopify / River / Midjourney — Related through Simon’s commentary on public-by-default systems and organizational learning loops.
- IBM / Granite 4.1 3B — Connected through Simon’s comparative SVG experiment with quantized model variants.
- Prompt injection / coding agents / agentic engineering / manual testing / automated tests — Broader themes Simon frequently touches that matter to product safety, reliability, and deployment strategy.
Newsletter Mentions (61)
“#10 📝 Simon Willison llm 0.32a2 - llm 0.32a2 adds several useful features, with a key change being support for OpenAI models using the /v1/responses endpoint so reasoning-capable models can interleave reasoning and tool calls; the release highlights summarized reasoning tokens displayed separately and introduces flags to hide reasoning if desired.”
#10 📝 Simon Willison llm 0.32a2 - llm 0.32a2 adds several useful features, with a key change being support for OpenAI models using the /v1/responses endpoint so reasoning-capable models can interleave reasoning and tool calls; the release highlights summarized reasoning tokens displayed separately and introduces flags to hide reasoning if desired.
“Simon highlights Tobias Lütke's description of Shopify's public, Slack-based coding agent 'River' that encourages learning by making work visible.”
#8 📝 Simon Willison Learning on the Shop floor - Simon highlights Tobias Lütke's description of Shopify's public, Slack-based coding agent 'River' that encourages learning by making work visible. He relates it to Midjourney's public Discord approach and suggests public workflows can accelerate learning.
“#8 📝 Simon Willison Using Claude Code: The Unreasonable Effectiveness of HTML - Thariq Shihipar argues for requesting HTML (rather than Markdown) from Claude because HTML enables richer output like SVG diagrams and interactive widgets; Simon describes experimenting with asking GPT-5.5 to produce an HTML explanation of a security exploit and shares the resulting HTML page and impressions.”
#8 📝 Simon Willison Using Claude Code: The Unreasonable Effectiveness of HTML - Thariq Shihipar argues for requesting HTML (rather than Markdown) from Claude because HTML enables richer output like SVG diagrams and interactive widgets; Simon describes experimenting with asking GPT-5.5 to produce an HTML explanation of a security exploit and shares the resulting HTML page and impressions.
“#3 📝 Simon Willison llm-gemini 0.31 - Release announcement for llm-gemini 0.31 noting that gemini-3.1-flash-lite is no longer a preview.”
Simon Willison is cited in two separate items, one about llm-gemini and another about Firefox hardening with Claude Mythos preview.
“Simon Willison Live blog: Code w/ Claude 2026 - Live notes from Anthropic’s Code w/ Claude event covering the morning keynote sessions.”
#19 📝 Simon Willison Live blog: Code w/ Claude 2026 - Live notes from Anthropic’s Code w/ Claude event covering the morning keynote sessions. The post is a live blog capturing highlights and observations from the event. #25 📝 Simon Willison Vibe coding and agentic engineering are getting closer than I’d like - A reflection on AI coding tools following a podcast conversation, highlighting concerns that vibe coding and agentic engineering are beginning to converge.
“#5 📝 Simon Willison Granite 4.1 3B SVG Pelican Gallery - Simon tried prompting different quantized variants of IBM's Granite 4.1 3B model to 'Generate an SVG of a pelican riding a bicycle' and published a gallery of the results.”
#4 𝕏 NVIDIA AI now offers end-to-end support in Megatron Core for training 30B-scale Kimi K2 and Qwen3 models with higher-order optimizers (Muon, MOP, REKLS), pushing efficiency on GB300 GPUs and NVL72 systems beyond standard data-parallel methods. #5 📝 Simon Willison Granite 4.1 3B SVG Pelican Gallery - Simon tried prompting different quantized variants of IBM's Granite 4.1 3B model to 'Generate an SVG of a pelican riding a bicycle' and published a gallery of the results. He found no clear relationship between model size and output quality — most results were poor. #6 📝 Anthropic News Building a new enterprise AI services company with Blackstone, Hellman & Friedman, and Goldman Sachs - Anthropic announced plans to build a new enterprise AI services company in partnership with Blackstone, Hellman & Friedman, and Goldman Sachs.
“Simon highlights migration tips, a Codex command to migrate projects, and a warning to treat GPT-5.5 as a new model family rather than a drop-in replacement.”
OpenAI Releases GPT-5.5 Prompting Guide #1 📝 Simon Willison GPT-5.5 prompting guide - OpenAI published prompting guidance for GPT-5.5, including a recommendation to send a short user-visible update before tool calls in multi-step tasks to improve UX. Simon highlights migration tips, a Codex command to migrate projects, and a warning to treat GPT-5.5 as a new model family rather than a drop-in replacement.
“Also covered by: @Simon Willison , @Cursor , @Aravind Srinivas”
#1 𝕏 Sam Altman announced that GPT-5.5 and GPT-5.5 Pro are now accessible via the API, enabling builders to integrate the latest model upgrades into their apps. Also covered by: @Simon Willison , @Cursor , @Aravind Srinivas
“Also covered by: @Peter Yang , @Simon Willison , @Sam Altman , @claire vo 🖤 , @Claire Vo , @OpenAI News #2 𝕏 Google AI launched Gemini 3.1 TTS with new inline audio tags (e.g., [whispers], [screams], [slow], [long pause]) that let you precisely control vocal style, pacing, and delivery.”
OpenAI Launches GPT-5.5 Model #1 📝 OpenAI News Introducing GPT-5.5 - OpenAI announces GPT-5.5, a new generation of its large language model series, highlighting improved capabilities and performance. The post introduces the model and likely outlines product implications and availability. Also covered by: @Peter Yang , @Simon Willison , @Sam Altman , @claire vo 🖤 , @Claire Vo , @OpenAI News #2 𝕏 Google AI launched Gemini 3.1 TTS with new inline audio tags (e.g., [whispers], [screams], [slow], [long pause]) that let you precisely control vocal style, pacing, and delivery.
“Simon adapted LlamaIndex's LiteParse (a Node.js CLI for extracting text from PDFs) to run entirely in the browser using the same libraries.”
#15 📝 Simon Willison Extract PDF text in your browser with LiteParse for the web - Simon adapted LlamaIndex's LiteParse (a Node.js CLI for extracting text from PDFs) to run entirely in the browser using the same libraries. He explains the work and provides a longer write-up with details and examples.
Related
Anthropic’s coding-focused assistant/tool used for building and automating engineering workflows. The newsletter references it in both security and product-usage contexts.
AI company behind Claude and related developer tools. In this newsletter it is highlighted for internal use of Claude Code and for product expansion into legal workflows.
The company behind ChatGPT and Codex, highlighted for launching Daybreak and a new deployment subsidiary for enterprise AI. It is positioned here as a platform provider moving deeper into cyber defense and enterprise deployment.
Anthropic’s assistant/model family, referenced in enterprise deployment, managed agents, and coding workflows. For AI PMs, it is central to agentic product design and enterprise integration.
An AI coding assistant with agentic and fast modes for development workflows. The newsletter notes a new Fast mode for Claude Opus 4.7 in Cursor.
A creator and commentator who shares practical workflows for Claude Code and personal operating systems for agents. He appears here as a curator of implementation advice for AI builders.
An AI framework company focused on retrieval, indexing, and data tooling for LLM apps. Here it is credited with launching an open-source parsing server.
AI developer advocate and educator known for tutorials around Gemini and open-source AI tooling. He is referenced here for a guide to the Gemini Interactions API.
Product and growth writer/podcaster focused on startups and PM topics. He is cited here for commentary on Anthropic’s operating pace and PM compensation content.
OpenAI’s coding-focused model/tool referenced as part of Daybreak’s security platform. For AI PMs, it signals coding intelligence being applied to cyber defense workflows.
A software project/company referenced as the codebase Garry Tan worked in while fixing a Dockerfile PATH issue with AI-generated code.
Google’s frontier AI research organization. The newsletter references it for launching interactive experiments in Google AI Studio.
A product lead associated here with Gemini API and AI Studio announcements. Known for shipping developer-facing AI product features.
A practitioner who used Claude and Cursor to generate a design system from GitHub repos. Relevant to PMs for rapid product and design-system iteration.
An AI researcher and founder known for practical prompting advice. Here he recommends ending prompts with HTML or slideshow formatting to get richer rendered outputs.
The company behind Gemini, referenced through a Gemini API quickstart guide. It is relevant for model access and developer onboarding.
Google’s AI model/product family, mentioned as one of the LLMs that names brands in category queries. In this newsletter it appears in the context of AI visibility and brand discovery.
AI researcher and educator known for practical machine learning content. In this newsletter he is credited with sharing a from-scratch Gemma 4 notebook on GitHub.
AI model family/company referenced as partnering with Fireworks AI to deploy closed-weight models in production.
Founder and CEO of Perplexity. He is mentioned here for technical commentary on GPU serving and MoE inference efficiency.
An open AI platform and ecosystem company focused on models, datasets, and infrastructure. The newsletter mentions both its infrastructure pitch and its dataset scale milestone.
Co-founder and CEO of Google DeepMind. He is mentioned here in relation to new funding for Isomorphic Labs and a Gemini-powered UI prototype.
Google Research/AI leader known for technical announcements around model deployment and infrastructure. Here, he is cited for announcing Gemini-powered translations in Google Search.
CEO of OpenAI, mentioned in connection with the launch of Daybreak and its cyber defense partnership invite. He is presented here as a spokesperson for OpenAI’s enterprise and security expansion.
CEO of Google and Alphabet. He is cited here as the announcer of Gemini Intelligence at Android Show I/O.
Autonomous or semi-autonomous systems that can plan and execute tasks using tools and models. The newsletter frames several product launches and startup strategies around agent-first workflows.
Anthropic Labs is mentioned as the organization where Henry Shi works with the founders. It appears as part of the credibility framing for the sponsored AI PM certification.
An AI development pattern where models act more like autonomous coding agents. The newsletter uses it to describe both NVIDIA Dynamo’s target workload and GPT-5.5/Codex improvements.
Anthropic’s latest Opus-class model release with a 1 million-token context window. It is positioned for long-context planning, coding, and agentic task execution.
A Claude model version referenced as part of a prompt-comparison analysis. It serves as one endpoint for examining changes in Anthropic’s system prompt evolution.
A Gemma model referenced alongside Multi-Token Prediction, with variants E2B/E4B. Important for PMs interested in open models and inference optimization.
A newer OpenAI model release with improved natural dialogue, longer context, and stronger tool use. It is discussed as a model now available in Cursor and chatprd.
A browser-related tool or workflow documented by LlamaIndex in a usage guide.
An AI coding assistant/orchestrator used to run stateful goal loops and automate coding workflows. It is presented here as a PM-relevant tool for agentic software development.
A Claude model variant referenced as the basis for Cursor’s Fast mode. It is presented as a higher-cost, faster option for coding tasks.
GPT-5.5 is a GPT model referenced as a writing/explaining assistant in the newsletter. It is used here to generate an HTML explanation of a security exploit.
Consumer technology company that builds iPhone, Mac, and Apple Intelligence features. In this newsletter it is referenced as partnering with Google for future Apple Intelligence capabilities.
A model-routing platform used to call multiple LLMs through a common interface. Here it is used to run four models in parallel for comparison and generation tasks.
Agents that perform coding tasks and can increasingly orchestrate adjacent workflows like design. The newsletter uses them as the execution layer for Design.md scripts.
A Gemini model variant that was noted as moving out of preview status.
A Claude model version referenced for more intelligent outputs with higher token usage. It is discussed alongside Opus 4.6 and effort settings for economical runs.
A Qwen model release referenced alongside Qwen3.6-Plus and integrated with opencode. It is one of the named models in the announcement.
A W3C-backed browser extension that exposes website functionality to MCP-capable agents. It lets developers register site functions as structured tools in the browser.
A browser automation protocol used here to let a Claude Code agent control Chrome programmatically.
A generative media company referenced as an example of a public Discord-based workflow. It is used here to support the idea that visible communities can accelerate learning and product adoption.
A Qwen model release with day-0 support for multimodal integration. The newsletter highlights its immediate compatibility with MLX-VLM for visual-language workflows.
Simon Willison’s command-line LLM tool for interacting with models and APIs. This release adds support for OpenAI’s Responses endpoint and better reasoning-token handling.
Technology company that offers the Granite family of models. In this newsletter it appears in relation to Simon Willison's prompting experiments with Granite 4.1 3B.
A collection of techniques and patterns for building agentic systems. The newsletter frames it as a guide page for AI builders.
A Google AI text-to-speech model with native multi-speaker dialogue support across many languages. It is positioned as part of the Gemini product family.
Google AI Edge Gallery is a Google tool for showcasing and running on-device AI experiences at the edge, including offline use cases.
A test-driven development pattern adapted for coding agents. It emphasizes an iterative failure/success loop that can make agentic coding more reliable.
A security risk pattern where AI agents have private data access, ingest untrusted content, and can exfiltrate data. For AI PMs, it is a key framework for designing safe agent features.
The practice of building software systems where agents plan and execute tasks with autonomy. The newsletter uses it in the context of anti-patterns and agent behavior management.
An ecommerce company referenced for its public, Slack-based coding agent River. The example is used to discuss how visible workflows can accelerate learning and adoption.
A quoted individual in a commentary about code quality incentives in AI systems. The newsletter uses him as the source of a viewpoint on maintainable code.
Attack technique where malicious prompts manipulate AI systems or agents. Here it is connected to a GitHub issue triage workflow exploit.
OpenAI leader and product/engineering voice associated here with confirming Codex’s unification with the main model. The newsletter cites him via Simon Willison’s note.
A model family from Google used as the base for TranslateGemma. It matters to PMs as an example of reusing a foundation model for a specialized, deployable product.
Armin Ronacher is a developer and writer who often explores AI tooling and infrastructure. In this issue he is credited with a piece on local models, inference engines, and serving ergonomics.
A Python library for working with LLM providers through an abstraction layer. The newsletter notes that API research is informing a major change to its provider abstraction.
A repository for researching LLM providers' HTTP APIs. It supports abstraction-layer decisions for developers building against multiple model providers.
A product and engineering concept describing the hidden cost of AI-accelerated development when teams lose shared understanding of the system. It reframes debt from code maintenance to team cognition and system comprehension.
A Codex-powered model release from OpenAI aimed at developers and product teams. The newsletter emphasizes its availability as a research preview and its high token throughput.
An open-weight multimodal model in Alibaba's Qwen3.5 series, aimed at agentic and vision-capable use cases. It is relevant to PMs evaluating model capabilities, openness, and deployment options.
Stay updated on Simon Willison
Get curated AI PM insights delivered daily — covering this and 1,000+ other sources.
Subscribe Free