OpenAI's ad platform has shifted from experiment to infrastructure in a matter of weeks. The company has moved from 200k up-front commitments to 30-50k monthly deals. Criteo and Smartly are now partners, and the ads business is already generating 100m ARR. Last week it launched an ads manager. This week it shipped a tracking pixel and an ads bot was spotted in the wild. The target is 2.5bn in ad revenue by the end of 2026 and 100bn by 2030.

The tracking pixel is the tell. You don't build a pixel unless you're serious about conversion attribution, something Facebook and Google have dominated for years. OpenAI says ads won't influence answers. I don't believe it. A 100bn revenue target moves that line.

The bit nobody's talking about is the psychology. Google users have 25 years of trained behaviour around ads. You know a sponsored result when you see one (although Google has done its best to blur the lines). ChatGPT users have none of that. They've spent three years being told this is an advisor, a personal assistant, an extension to your brain. It's closer to asking a friend than querying a database. When ads land inside that experience, disclosure won't protect the trust. A recommendation embedded in a paragraph of natural language feels like advice, even with a sponsored label.

Once users know ads exist in the system, the "is this genuine or paid for?" question attaches itself to every answer, not just the sponsored ones. That's the real risk. Once suspicion attaches to the response, the wall between ads and answers doesn't matter. It's already been breached in the user's head.

Google's early ads in the 2000s were text-only, clearly separated, and users still pushed back. It took years of gradual blurring before people stopped noticing. OpenAI doesn't have that runway. Zero to 2.5bn by end of 2026 isn't gradual. It's a step change, and the behavioural backlash will be proportional.

The question for every brand and every user is whether the trust survives the transition.

The question for investors is whether Sam Altman is the right man to lead the company into the next phase of its journey and blockbuster IPO as the OpenAI CEO has some clear conflicts of interest, which surely clouds his decision making.

While OpenAI is figuring out how to monetise its users, the AI search tooling category is heating up fast. Bluefish just raised $43 million. Peec raised $29 million. Profound is sitting on $155 million. Obsero is still bootstrapped in a hugely competitive category that is continuing to grow.

Money like this in a category tells you the AI search discovery problem is real. Nobody's writing 43m cheques for a fad. But raised capital changes how you build. VC-funded tools optimise for scale, board metrics, enterprise sales cycles. Bootstrapped means you build what customers need. That's a different lens, and I think it's the right one for us at this stage of our journey.

We also changed our structure this month and defined roles more clearly. Mike and I dropped our generic 'co-founder' titles and became CPO and CCO. I'm running product. Mike's running customer. Co-founder is what you call yourselves when you're still figuring out what the company is. CPO and CCO is what you call yourselves when you know.

That clarity showed up in what we shipped next. We launched Obsero Local - city and postcode-level tracking across ChatGPT, Perplexity, Gemini, Google AI Mode and AI Overviews.

Everyone in this category is tracking the same five platforms at a national level. Nobody was asking what happens when you zoom in. But the models ARE answering locally. Ask ChatGPT "best Italian restaurant near me" in London versus Leeds and you get different answers. Almost nobody is measuring this.

Although it was a feature we discussed at the start of our journey, Local came straight out of customer conversations Mike was having. Multi-location brands saying "our national data looks fine, but what's happening in Soho or Salford?" That's the CPO and CCO working in sync. Mike hears the problem. I ship the answer.

Christian J Ward, Chief Data Officer at Yext, told me on Franc Talking that his team analysed 1.8 million questions and 7 million citations to understand what location signals do to AI answers. His conclusion was blunt: location signals matter enormously. Obsero agrees.

Local was hugely important in organic search and something we tackled significantly when I was leading the organic function at Cazoo. Brands that understood their customer and their category were able to win across Google Maps and the local SERPs. Same thing's about to happen in AI search.

If you run a multi-location brand, pick your five most commercially important locations and run the same five prompts in each. If the answers vary wildly, you've got a content and entity problem at the local level. If your brand appears in zero of them, that's the bigger problem, and it's the one most multi-location brands don't know they have.

Interested in trying this out? Drop me an email on [email protected].

Quick disclosure. I use Claude every single day. Obsero, client work, the newsletter, this edition. It's not a tool I dabble with, it's how I work. So this isn't a neutral take, but I'd rather tell you what's actually useful than sell you on the hype.

Anthropic rolled out Claude Opus 4.7 this week. Annualised revenue jumped from 9bn at end of 2025 to 30bn by end of March, driven largely by enterprise adoption. 80% of their revenue is business customers. Eight of the Fortune 10 use Claude. VCs are circling at 800bn valuations with an IPO reportedly pencilled for October. None of that matters to you as a marketer. What matters is what you can build with it in your office/workspace.

The thing that's changed in the last six months is MCPs. Model Context Protocol. Claude can now plug directly into your tools. beehiiv, Notion, Google Drive, PostHog, Slack, Canva, Framer. Not in a "copy and paste an API key" way. In a "Claude, pull my last five newsletter opens and tell me what's working" way. This is how I have been embracing Claude.

I've got Claude connected to my beehiiv account. I can ask which of my last ten subject lines had the highest open rate and what they have in common, and it just does it. That's a workflow that used to be export CSV, pivot table, stare at it, write something. Now it's a single conversation.

And then this week, Anthropic shipped Claude Design. If MCPs are Claude reaching into your tools, Claude Design is Claude reaching into your design stack.

What it actually does is insane.

Slides, video, wireframes, interactive prototypes, landing pages, pitch decks, one-pagers. You describe it, it builds it. It reads your codebase and design files on the way in, builds a design system off them, then applies it automatically to everything it makes. So your output looks like your brand, not like a generic AI slide deck. Export straight to Canva, PDF, PPTX or standalone HTML. If it's ready to build, hand it to Claude Code and it ships.

That's the bit that breaks my head a little. The line between "I need a designer" and "I need Claude for twenty minutes" is moving super fast. For a founder or a marketer without a design team, this is essential. And it's not a separate product. It's the same Claude, the same chat, just with a design canvas attached (although I can’t find a shortcut on the desktop app and am using it in the browser). Which is exactly the Anthropic pattern. They don't launch twelve things, they launch one thing that does twelve jobs.

Pick one tool you use every day that has an MCP. Connect it to Claude. Don't try to automate everything. Pick the one reporting task you hate doing on a Friday afternoon and get Claude to do it. Set it up once, let it run, and use it for thinking, not just doing. The best use of Claude isn't "write this for me." It's "challenge my thinking on this." That's where you get the real leverage.

Anthropic versus OpenAI isn't just a VC story. It's a question about which AI you trust to sit inside your workflow.

OpenAI is launching an ads manager and chasing consumer reach. Anthropic is winning enterprise. ChatGPT is winning the search bar. Claude is winning the workflow. Your visibility strategy needs to cover both. Your productivity strategy definitely leans one way.

In 2026, it’s easier than ever to start a newsletter about a topic you love, grow your audience, and increase revenue.

Whether you’re in B2B SaaS, FinTech or want a side hustle, consider starting a newsletter today and beehiiv is the perfect partner for this.

Use my link here and start your 14-day trial today. If you sign up, you get 20% off your subscription fee for 3-months!