jc.Résumé PDF ↗
← All briefings

· The Intelligence Brief

The Intelligence Brief

AI gets cheaper, capital gets picky, and health records plug into ChatGPT — today the story is AI maturing from novelty to infrastructure, and what that means for your rate card.

The briefing

Cheap tokens change the business model

Token prices hitting record lows is the quiet headline of the day. When inference costs almost nothing, AI-powered products stop being expensive demos and become viable at Nigerian price points. If you've been shelving an AI product idea over API bills, revisit the maths today. The value is migrating from access to models toward the application layer — products, workflows, and domain expertise.

Regulated data is the new frontier

OpenAI shipping read-only Epic integration for patient records is significant beyond healthcare: it shows AI vendors are learning to work inside strict compliance regimes rather than around them. This opens a path for AI work in finance, insurance, and health tech — sectors where African developers can charge premium rates for combining AI with compliance knowledge. Second-order effect: 'prompt engineer' keeps losing value; 'AI + regulated domain' keeps gaining it.

Capital and competition both squeezing

Anthropic's looming mega-IPO is crowding out other listings and signalling that investors now want proven AI companies, not promises. Meanwhile Google is preparing a Gemini push aimed squarely at coding leadership, threatening OpenAI and Anthropic's grip. For freelancers and founders, this means clients will increasingly ask which model you're building on — and the honest answer should be 'whichever wins next quarter,' so architect for portability.

The stories

  1. TechCrunch

    ChatGPT Health adds Epic integration for clinicians to import patient data

    Clinicians can now pull patient records into ChatGPT — a template for AI moving into regulated, high-stakes data work, exactly where trust and compliance skills pay best.

  2. CNBC

    AI token prices are hitting new record lows

    Falling token prices mean inference costs are collapsing, so the margin is shifting from running models to what you build on top of them.

  3. Bloomberg

    Anthropic’s mega-IPO plan looms over packed US listing calendar

    Anthropic's IPO is soaking up investor attention, tightening funding for everyone else — build lean because the fundraising window just narrowed.

  4. The Times of India

    US to press G20 on light-touch AI regulation

    The US pushing light-touch AI rules at G20 signals a friendlier regulatory climate for builders, at least in markets that matter to your clients.

  5. India Today

    Google almost ready to launch new Gemini AI model, may beat Anthropic and OpenAI in coding this time

    Google's next Gemini may take the coding crown, meaning the 'best AI coder' title is now a rotating seat — don't marry one toolchain.

Three angles

Build to learn

Generic 'AI skills' are commoditising — specialise or sink

Between dirt-cheap tokens and Google's Gemini gunning for coding supremacy, basic AI-assisted coding is becoming table stakes everyone has. The workflow under pressure is undifferentiated coding and simple prompt-based services. Move up: pick a regulated domain like health or fintech, learn the compliance side, and sell AI plus domain depth.

Build to sell

Collapsing token prices make USD-priced AI products viable now

With token prices at record lows, an AI product that cost $500/month in API bills last year might cost $50 today. Build a niche tool for US or European clients — the same codebase, priced in dollars, insulates you from naira swings. The cost barrier to entering the AI product business effectively disappeared this quarter.

Build early

Quiet founders are betting on portability, not loyalty

Sharp Lagos operators aren't arguing about which model is best — they're building thin abstraction layers so their products swap between OpenAI, Anthropic, and Gemini in an afternoon. With Google possibly taking the coding crown and token costs near zero, the smart money is on infrastructure-agnostic products and compliance-ready data pipelines, not model fandom.