jc.Résumé PDF ↗
← All briefings

· The Intelligence Brief

The Intelligence Brief

AI's biggest players wobbled, OpenAI played politics with a billion dollars, and agents that run your whole computer arrived — the ground is shifting under everyone building on these tools.

The briefing

Concentration risk is now real

Thursday's simultaneous outage of ChatGPT, Claude and Grok showed the AI layer has shared fragility — if you build on one provider, you build on their bad days. OpenAI walking away from Cursor over a Musk feud reinforces the point: platforms can vanish for reasons that have nothing to do with you. Today, multi-provider fallback stopped being paranoia and became basic engineering hygiene.

Generic output is losing value fast

The AI-menu story and GPT-6 Astra's computer-use pitch point the same direction: the cheap, templated stuff is getting automated or rejected. What remains scarce is judgment — knowing what's good, not just generating what's plausible. If your work is mostly assembling standard outputs, both customers and agents are coming for it.

Foundation-model economics are getting political

Anthropic eyeing a $2 trillion IPO with external trustees means its mission-driven pricing and access policies will face shareholder pressure. Combine that with OpenAI's billion-dollar grudge match and the picture is clear: the models you depend on are being steered by power struggles and capital markets, not just research. Budget and architect accordingly.

The stories

  1. TechCrunch

    The sameness problem behind those unappetizing AI-generated menus | TechCrunch

    AI slop is now a trust signal — customers can smell generic output, which means human taste and editing are becoming the paid differentiator.

  2. TechStartups.com

    Widespread AI outage hits ChatGPT, Claude and Grok at the same time

    ChatGPT, Claude and Grok failing simultaneously proves your workflow needs a fallback model and local copies of critical prompts.

  3. WIRED

    OpenAI Cut Off a Billion-Dollar Customer to Avoid Elon Musk

    OpenAI abandoned $1B+ in Cursor revenue over a Musk conflict — vendor politics, not just tech, can kill the tools your stack depends on.

  4. The Verge

    Watch GPT-6 Astra use your computer for you.

    GPT-6 Astra pitching full computer control means routine UI work — data entry, slide decks, form filling — is next in line for automation.

  5. Ars Technica

    Anthropic’s $2 trillion IPO puts powerful external trustees in spotlight

    Anthropic's unusual trustee structure faces public-market pressure, so Claude's pricing and priorities could shift once shareholders demand returns.

Three angles

Build to learn

Computer-use agents target routine UI work

GPT-6 Astra's demo of assembling presentations and operating desktops puts data entry, basic VA tasks and slide production directly in the automation path — work many Nigerian freelancers bill for. The move: shift from executing tasks to designing and supervising the workflows agents run, and build the fallback skills today's outage proved you'll need.

Build to sell

Be the human taste layer clients pay USD for

The AI-menu story is your pitch: businesses globally are discovering customers reject obvious AI output and will pay for humans who edit, localize and quality-check it. Offer 'AI content refinement' or 'human-in-the-loop' packages on Upwork and Fiverr today — it's a USD skill that yesterday's news just made more sellable, and it survives the naira.

Build early

Quiet Lagos shops are de-risking their model stacks

Senior operators watched the same outage and are quietly wiring multi-provider routing — Claude as fallback for GPT, local model caches for critical prompts — before clients notice. The sharper ones are also reassessing Cursor reliance after the OpenAI fallout. Nobody's announcing it; they're just not going down when the next outage hits.