OpenAI and Anthropic Are Now Consulting Firms. What That Tells Small Businesses.

In the first week of May 2026, two announcements landed almost on top of each other. OpenAI confirmed it was launching OpenAI Deployment Co., a consulting arm to help organisations build and deploy AI systems, and said it would acquire applied-AI firm Tomoro. Days earlier, Anthropic had spun up a joint venture with several Wall Street financial-services firms to sell and install AI tools to enterprises. Two of the largest AI model vendors on the planet decided, within a week, that they need to be in the implementation business. That's a tell.

What it means for SMBs

Model vendors have long pitched the same story: "Our model is smart enough that you can just plug it in and it'll work." The fact that the same vendors are now hiring (or buying) consulting capability means that story has finally bumped into reality. The hardest part of useful AI deployment was never the model — it was the integration, the data hygiene, the change management, the security boundaries. The vendors know it. Their customers have been telling them for two years.

For a small business, this matters in two ways. First, it confirms a pattern we've been seeing on the ground for months: the businesses getting real value out of AI aren't the ones who picked the smartest model — they're the ones who scoped the problem narrowly, wired the tool into existing workflows, and trained the team to use it. Smart model selection is maybe 20% of the outcome.

Second, the consulting arms of OpenAI and Anthropic will overwhelmingly chase Fortune 500 contracts. Their pricing and engagement model is built for that. Small and mid-sized businesses still have to figure out the integration problem somewhere else, with someone whose minimums and approach actually fit a smaller engagement.

The same week also brought a notable signal from Ramp's business AI index: Anthropic's Claude overtook OpenAI's ChatGPT in U.S. business adoption for the first time, with Claude reaching roughly 34% adoption versus ChatGPT's 32%. Both share continues to grow — just not at the same rate. For SMBs that means the "default" enterprise AI choice is genuinely competitive now, and locking in to one vendor is a less defensible decision than it was a year ago.

What to do about it

A few practical moves while the dust settles:

• Don't pick a single AI vendor and standardise on it yet. Both Anthropic and OpenAI now have business plans with meaningful differentiation. Run both in parallel for at least one quarter before standardising. The cost difference is rounding error compared to the cost of picking wrong.
• Inventory where AI is already in your business — including shadow usage. Most SMBs we audit find 3-5 AI tools their team is paying for personally, with company data going into them. That's a security exposure and a procurement opportunity.
• Match the deployment effort to the value. Customer-facing agents need real engineering. Internal drafting tools need almost none. Don't run an internal experiment with enterprise process.

Where we come in

The vendors' consulting arms exist because integration is hard. We do exactly that work for businesses too small to be on Anthropic's or OpenAI's enterprise call sheet — vendor-neutral, scoped to your actual problem, priced for an SMB engagement. If you'd like a second opinion on your current AI stack or help shipping the next deployment, reach out. Our AI consulting and custom AI agents work is built for this kind of thing.