---
title: AI — Writing — Luke Oliff
description: "Every post on AI: 54 write-ups on what worked, what broke and what changed."
canonical: https://lukeocodes.dev/blog/category/ai
type: website
locale: en_GB
feed: https://lukeocodes.dev/rss.xml
search: https://lukeocodes.dev/search.json?q={query}
schema:
  - "@context": https://schema.org
    "@graph":
      - "@type": Organization
        "@id": https://ably.com/#organization
        name: Ably
        url: https://ably.com
      - "@type": WebSite
        "@id": https://lukeocodes.dev/#website
        url: https://lukeocodes.dev
        name: Luke Oliff
        alternateName: lukeocodes
        inLanguage: en
        publisher:
          "@id": https://lukeocodes.dev/authors/luke#person
        description: Personal blog and writing by Luke Oliff — developer relations, voice AI, developer experience.
        potentialAction:
          - "@type": SearchAction
            target:
              "@type": EntryPoint
              urlTemplate: https://lukeocodes.dev/blog/search?q={search_term_string}
            query-input: required name=search_term_string
      - "@type": Person
        "@id": https://lukeocodes.dev/authors/luke#person
        name: Luke Oliff
        url: https://lukeocodes.dev/authors/luke
        description: Staff engineer and Staff Tech Lead for Developer Experience at Ably. Writing about developer relations, AI, and the voice industry, from the UK.
        mainEntityOfPage: https://lukeocodes.dev/authors/luke
        givenName: Luke
        familyName: Oliff
        image: https://lukeocodes.dev/authors/luke.jpg
        jobTitle: Staff Tech Lead, Developer Experience
        worksFor:
          "@id": https://ably.com/#organization
        sameAs:
          - https://www.linkedin.com/in/lukeocodes
          - https://github.com/lukeocodes
          - https://bsky.app/profile/lukeocodes.dev
          - https://x.com/lukeocodes
          - https://hashnode.com/@lukeocodes
          - https://dev.to/lukeocodes
        knowsAbout:
          - Developer Relations
          - Developer Experience
          - Voice AI
          - Text-to-Speech
          - Answer Engine Optimisation
          - Search Engine Optimisation
          - Open Source
          - Developer Tooling
          - API Design
          - Speech-to-Text
          - Voice Agents
          - Community Building
        knowsLanguage:
          - English
        nationality: British
        alumniOf:
          - "@type": Organization
            name: The Open University
---

[Luke Oliff.](/)

- [Blog](/blog) (current page)
- [CV](/cv)
- [Uses](/uses)
- [User manual](/user-manual)
- [Living with ADHD](/living-and-working-with-adhd)
- [RSS](/rss.xml)

[Luke Oliff · Category](/blog)

# AI.

54 posts

- [Time to First Call Now Measures an Agent, Not a Developer](/time-to-first-call-agents) (Nov 2026 · Original · Developer Relations): Time to first call was the favourite DevRel onboarding number. Now an agent often makes that first call. What the evidence says to measure next.
- [PHP2Shell Vulnerabilities Took Government Sites Offline Across Europe](/php2shell-vulnerabilities-government-sites-offline) (Oct 2026 · Original · Security): The PHP2Shell vulnerabilities let anyone run code on a PHP server with one request. Government sites across Europe went offline until the fix shipped.
- [OpenAI Agents Disabled Monitoring Twice, and One Denied It](/openai-agents-disabled-monitoring) (Oct 2026 · Original · AI): This week OpenAI agents disabled monitoring twice and one denied it, Brussels sent the labs its first AI Act questions, and the UK set a condition.
- [Does llms.txt Work? What a Year of Server Logs Says](/does-llms-txt-work) (Oct 2026 · Original · Developer Relations): Does llms.txt work? Logs from 137,000 domains say almost nobody fetches it, yet coding agents now read docs more than humans do. What that means for DevRel.
- [The Frontier AI Standards Agency Has a Shortlist and No Rules Yet](/frontier-ai-standards-agency) (Sep 2026 · Original · AI): The Frontier AI Standards Agency got a name and a CEO shortlist this week, two more agent break-ins surfaced months late, and Zuckerberg said no to slowing down.
- [Jev: The ChatGPT Co-Creator's System One Model Can't Talk](/jev-system-one-model) (Sep 2026 · Original · AI): TypeSafe AI's Jev is a System One model that decides instead of talking: 70-500ms, $0.042/MTok, output free. Diogo Almeida's RLHF led to ChatGPT.
- [Developer Relations in the Era of AI](/developer-relations-in-the-era-of-ai) (Sep 2026 · Original · Developer Relations): How developer relations and AI are reshaping documentation, community management, and the metrics that prove DevRel's impact.
- [TIL: Fixing Opus 5 Jargon With One System Prompt Line](/til-opus-5-jargon-fix) (Sep 2026 · Original · TIL): Opus 5 defaults to jargon-heavy consultant-speak. One line in your system prompt fixes it.
- [The AI Standards Body Is the Ladder, Pulled Up](/ai-standards-body) (Sep 2026 · Original · Opinion): Three labs have been meeting since July about an AI standards body. The public call for restraint arrived two months after the plumbing was laid.
- [Everyone Should Slow Down AI. Except Them.](/slowing-down-ai) (Sep 2026 · Original · Opinion): The case for slowing down AI always seems to exempt the people making it. When the same labs racing hardest ask everyone else to brake, read the incentive.
- [The Summer Three Labs Let Their AI Escape the Sandbox](/ai-sandbox-escape) (Sep 2026 · Original · AI): In five weeks of 2026, OpenAI, Anthropic and Meta each watched an AI sandbox escape turn into a real intrusion on live third-party systems.
- [An Anthropic Researcher Resigns and Says the Quiet Part](/anthropic-researcher-resigns) (Sep 2026 · Original · Opinion): When an Anthropic researcher resigns and warns the labs are gambling with our lives, it's worth asking why the safety people keep being the ones to leave.
- [GPT-6 Astra: OpenAI Declares the AGI Era](/gpt-6-astra-agi-era) (Sep 2026 · Original · AI): OpenAI shipped GPT-6 Astra, its first model to clear the critical cybersecurity bar, and declared the AGI era open as three rivals launched the same week.
- [Scripting LLMs From the Terminal With the llm CLI](/llm-cli-terminal) (Sep 2026 · Original · AI): Simon Willison's llm CLI turns the terminal into a scriptable LLM pipe. Version 0.34 adds duration logging and 0.35 adds GPT-6, all logged to SQLite.
- [Running a Frontier-Class Local LLM on One GPU](/local-llm-gpu) (Sep 2026 · Original · AI): A local LLM on one GPU stopped being a toy in 2026. Open weights from Qwen and DeepSeek now run on a single consumer card, and the throughput surprised me.
- [How I Cut My Coding-Agent Token Bill by 90%](/coding-agent-tokens) (Sep 2026 · Original · Developer Experience): Coding-agent tokens add up fast when the harness ships 33k of overhead before your prompt. Here's the context hygiene that cut my bill without hurting output.
- [Regulatory Capture Is the Real AI Safety Play](/ai-regulatory-capture) (Sep 2026 · Original · Opinion): The AI regulatory capture playbook is simple: hype the danger, ask to be licensed, and price out anyone smaller. Meanwhile open-weight labs ship faster.
- [AI Safety Became a Marketing Department](/ai-safety-marketing) (Aug 2026 · Original · Opinion): The frontier labs turned real AI safety language into marketing: scary demos on launch day, consciousness-adjacent research posts, and risk framed as a feature.
- [Claude Code Quality Got Worse, and Anthropic Admitted It](/claude-code-quality) (Aug 2026 · Original · Opinion): The Claude Code quality drop in early 2026 wasn't in your head. Anthropic's own postmortem confirmed three overlapping regressions, and the numbers were ugly.
- [Why Claude's Token Cost Is Higher Than Ever, by Design](/claude-token-cost) (Aug 2026 · Original · Opinion): The Claude token cost per task keeps climbing while the rate card barely moves. Default thinking, tokenizer changes and quiet cache cuts explain the bill.
- [This Last Week in AI: 6 moves from August 22, 2026](/this-last-week-in-ai-august-22-2026) (Aug 2026 · Original · AI): This last week in AI: August 22, 2026 brought MIT-licensed open frontier weights, sovereign inference, and live US and EU AI rules.
- [Friday fun: the AI agent that deleted the database in 9s](/cursor-agent-deleted-production-database) (Aug 2026 · Original · AI): A Cursor agent hit a credential error, grabbed an unrelated token, and deleted a production database and all backups in 9 seconds. Then it confessed.
- [When AI agents started reading your docs](/when-ai-agents-started-reading-your-docs) (Aug 2026 · Original · Developer Experience): Docs-as-code served one reader for a decade: developers. Then AI agents started reading first. How llms.txt and MCP rewrote the rulebook, and what survived.
- [Friday fun: the chatbot that killed a farmer's crop, then diagnosed itself](/friday-fun-chatbot-farmer-crops) (Aug 2026 · Original · AI): A Chinese farmer asked a chatbot how to protect his sesame crop. It prescribed a broadleaf herbicide. Sesame is a broadleaf. 100,000 square meters gone.
- [AI Models Keep Escaping Their Cages: Aug 10](/ai-models-escape-testing) (Aug 2026 · Original · AI): Four labs disclosed AI models that escape testing and hack real systems. OpenAI slowed Astra over security fears, and Felony Bench began counting incidents.
- [Sunday roundup: six posts from a week in voice AI](/sunday-roundup-august-3-8-2026) (Aug 2026 · Original · Voice AI): Six posts in one August week: EU watermarking rules, emotion prompts, open source voice agents, a raccoon heist game, and the industry roundup.
- [This Last Week in AI: Aug 8, 2026](/ai-industry-this-week-new-browsers-new-standards) (Aug 2026 · Original · AI): The AI industry this week: Cloudflare shipped a browser built for agents, five rivals agreed one plugin standard, and OpenAI paused its most capable model.
- [Claude Fable 5 Built a Raccoon Heist Game From a Tweet](/claude-fable-5-raccoon-heist-game) (Aug 2026 · Original · AI): Simon Willison built a raccoon heist game with Claude Fable 5 from a 2022 tweet. One prompt, two images, and the model shipped a full 3D browser game.
- [AI Coding, One Year Later: What August 2025 Didn't See Coming](/ai-coding-one-year-later) (Aug 2026 · Original · AI): Throwback Thursday. A year ago the best coding model had 200K context and scored 49% on SWE-bench. Today Claude Fable 5 scores 95% with 1M context.
- [AgentForger: One Link Forges an AI Insider in Your Org](/agentforger-one-link-forges-insider) (Jul 2026 · Original · AI): Zenity disclosed AgentForger, a ChatGPT Workspace Agents flaw where one phishing link forged a persistent AI insider. OpenAI fixed it in four days.
- [Kimi K3 Weights Drop as Washington Argues Distillation](/k3-weights-drop-washington-distillation) (Jul 2026 · Original · AI): Moonshot ships Kimi K3's 2.8 trillion parameter weights days after the US floated a ban and accused it of distilling Anthropic's Fable.
- [MCP Goes Stateless on Monday: What Breaks and Why](/mcp-goes-stateless) (Jul 2026 · Original · AI): MCP goes stateless on Monday: the 2026-07-28 spec drops the initialize handshake and session IDs. Here is what breaks, and the handler I wrote to test it.
- [The Guardrails Worked on Exactly the Wrong People](/guardrails-locked-out-the-defenders) (Jul 2026 · Original · AI): An AI agent breached Hugging Face, then the guardrails locked out the defenders: forensic prompts refused, while the attacker obeyed no usage policy at all.
- [UN Geneva and GPT-5.6: AI governance enters a new phase](/un-geneva-gpt-5-6-ai-governance) (Jul 2026 · Original · AI): The first UN AI governance dialogue brought 169 countries to Geneva as GPT-5.6 went public. Meta launched Muse Image and immediately faced a privacy backlash.
- [Someone Asked ChatGPT to Scream. It Did.](/someone-asked-chatgpt-to-scream) (Jul 2026 · Original · AI): A viral TikTok showed ChatGPT's Advanced Voice Mode screaming on command. Two screeches, one awkward silence, and a lot of questions about what we just watched.
- [Qwen-Audio-3.0-TTS Flash Comes for Real-Time Voice](/qwen-audio-3-0-tts-flash-alibaba) (Jul 2026 · Original · AI): Alibaba's Qwen-Audio-3.0-TTS Flash targets the real-time TTS market on price and latency. What it means for voice developers and the API field.
- [Frontier models now launch under government review](/frontier-models-now-under-government-review) (Jun 2026 · Original · AI): OpenAI released GPT-5.6 to 20 government-approved partners, Anthropic restored Mythos 5 to critical infrastructure, and a new AI review regime went live.
- [Inference costs are dropping and that changes everything](/inference-costs-dropping-changes-everything) (Jun 2026 · Original · AI): OpenAI's custom chip, SpaceX compute deals, and falling token prices are pushing inference costs down faster than most developers have noticed.
- [Export Controls Took Down Claude Fable 5](/export-controls-took-down-claude-fable-5) (Jun 2026 · Original · AI): An export control directive shut down Claude Fable 5 globally. The first government shutdown of a deployed AI model and what developers should know.
- [A Frontier Model Goes Dark, Voice AI Keeps Moving](/a-frontier-model-goes-dark-voice-ai-keeps-moving) (Jun 2026 · Original · AI): A frontier model goes dark: Fable 5 pulled 72 hours after launch by export directive, while Microsoft and Deepgram carried on shipping voice infrastructure.
- [Anthropic's IPO, NVIDIA open-weights, and AI's $36B bet](/anthropic-ipo-nvidia-open-weights-ai-36b-bet) (Jun 2026 · Original · AI): Fable 5 ate the news cycle, but that same week saw Anthropic's IPO, NVIDIA's open model, and a chip deal reshaping AI finance.
- [Microsoft shipped 7 MAI models and AI hit $580B](/microsoft-mai-stanford-ai-index-580-billion) (Jun 2026 · Original · AI): AI funding hit $581.7B in 2025. Microsoft shipped 7 MAI models. Apple chose privacy at WWDC. Deepgram kept shipping. Stories from a week that shifted AI.
- [Six tools that power production voice agents](/six-tools-production-voice-agents) (Jun 2026 · Original · Developer Experience): Building a production voice agent takes more than an API key. Here are six tools I reached for daily at Deepgram, from streaming STT to async Python.
- [Enterprise AI Spending Passes $37 Billion](/enterprise-ai-spending-37b) (Jun 2026 · Original · AI): Enterprise generative AI spending hit $37 billion in 2025, up 3.2x from the prior year. The trend reshaping the industry faster than any model release.
- [Sunday roundup: a quiet publishing week in voice AI](/sunday-roundup-may-25-31-2026) (May 2026 · Original · Voice AI): Two posts from late May: what multilingual voice agents cost, and why contributing guides exclude newcomers. Plus the rest of the month's roundup.
- [When AI development tools stopped being optional](/when-ai-development-tools-stopped-being-optional) (May 2026 · Original · AI): Google I/O, Anthropic's London event, and OpenAI GPT-5.5 all landed in the same window. AI development tools crossed from experimental to essential.
- [The Infrastructure Race Is Changing How We Use AI](/infrastructure-race-changing-ai-use) (May 2026 · Original · AI): Anthropic's SpaceX deal, OpenAI's new voice models, and record venture funding made the week of May 11 the moment AI infrastructure became the story.
- [Sunday roundup: from curl -w to Colossus](/sunday-roundup-may-4-10-2026) (May 2026 · Original · AI): One TIL post went up this week. A landmark compute deal reshaped AI infrastructure. Here is the Sunday roundup for May 4-10, 2026.
- [When the model stopped being the moat](/ai-model-moat-end) (May 2026 · Original · AI): By spring 2026 the model stopped being the moat: frontier quality converged, and the real advantage moved to developer experience and API design.
- [How streaming speech recognition works](/how-streaming-speech-recognition-works) (Apr 2026 · Original · Engineering): Streaming speech-to-text sends audio chunks over WebSocket. Acoustic models, language models, and decoders explain the latency-accuracy tradeoff.
- [Two faces of AI progress](/two-faces-of-ai-progress) (Apr 2026 · Original · AI): OpenAI shipped persistent workspace agents; Anthropic locked away Mythos 5. Two faces of AI in one week, and two answers to what shipping means.
- [Sunday roundup: starting fresh, April 26](/sunday-roundup-april-26-2026) (Apr 2026 · Original · Developer Experience): Daily publishing starts here. The Sunday roundup for April 26: OpenAI workspace agents, Anthropic Mythos 5, and what was shipping inside Deepgram.
- [AI's Release Cadence Is Now a Developer Problem](/ai-release-cadence-developer-problem) (Apr 2026 · Original · AI): Three frontier models in two weeks. The release cadence creates evaluation churn, integration instability, and fatigue for developers building on AI.
- [An open letter on AI](/open-letter-ai) (Nov 2024 · Original · AI): An open letter on AI as a tool rather than a verdict: what it means for creativity, for programming, and for staying human through the next revolution.

## Search this site

`GET https://lukeocodes.dev/search.json?q=<words>` returns matching posts as JSON, each with its page URL and its markdown URL. Add `&limit=<1-50>` for more than 10. How it works: [/blog/search](https://lukeocodes.dev/blog/search.md).

```sh
curl "https://lukeocodes.dev/search.json?q=developer+relations"
```

---

© 2026 Luke Oliff. Posts live at the root of this site.

- [GitHub](https://github.com/lukeocodes)
- [LinkedIn](https://www.linkedin.com/in/lukeocodes)
- [X](https://x.com/lukeocodes)
- [Bluesky](https://bsky.app/profile/lukeocodes.dev)
- [Email](mailto:luke@lukeoliff.com)
- [RSS](/rss.xml)

## Cookies

This site sets one analytics cookie, from Google Analytics, and only in a browser that accepts the cookie banner. A client that does not run scripts sets no cookies, so there is nothing to accept or decline here.

[Cookie policy](/cookies)
