Estimated read time: 7 minutes
Every so often a company tells you exactly where the money is by where it puts its own. This week, that company was Anthropic, and the number was $1.5 billion.
On July 15, TechCrunch reported that Anthropic, alongside Blackstone, Goldman Sachs, and Hellman & Friedman, has quietly built a company called Ode. Its job is not to build a smarter AI. Its job is to walk into other companies’ offices and make the AI they already own actually work. OpenAI built a nearly identical business with the blandest possible name, The Deployment Company. Deloitte and Accenture are scrambling to do the same thing.
Read that lineup again. The most valuable AI labs on earth just decided the next fortune is not in better models. It is in getting ordinary businesses to use the ones we already have. If you run a small business, that is the best news you will read all year, and almost nobody is framing it that way.
Table of Contents
What Anthropic actually did
Ode started as a problem Blackstone kept running into. As it pushed AI across the companies it owns, it hired big consulting firms and small specialist shops to do the actual installation work, and one small shop, an engineering startup called Fractional AI, kept outperforming everyone. So the new venture simply bought it. Fractional walked away from an eleven-month partnership with OpenAI, became the core of Ode, and now employs around 100 engineers who plug directly into Anthropic’s applied AI team.
The company operates on what its leaders call a “Claude-first” principle, meaning it reaches for Anthropic’s tools first and rival products only when it has to. And its ambitions are not small. “It’s pretty easy to imagine this as a trillion-dollar company someday if we execute well,” Ode CEO Chris Taylor told TechCrunch. A trillion-dollar company whose entire product is helping other people use software that already exists.
That is the part worth sitting with. Nobody is paying Ode for a new model. They are paying it for the boring, unglamorous, weirdly hard work of connecting a model to a real business process, and paying enough that Blackstone, Goldman, and Anthropic all wanted a piece.
The quiet admission in the headline
For two years the entire AI conversation has been a horsepower contest. Whose model scores higher. Whose context window is longer. Which lab shipped the newest number after the decimal point. Ode is a bet that none of that is where the value lives anymore.
Eddie Siegel, Ode’s chief technologist, put it about as bluntly as a technologist can: “Model selection matters, but it’s not where the majority of calories are spent. It’s like the choice of programming language when you build a piece of software.” In other words, which AI you pick is a footnote. The engineering around it, the judgment about which process to rewire and how, is the whole game.
That reframes the last two years. The reason most companies have spent money on AI and felt very little in return is not that they picked the wrong model. It is that they bolted a brilliant tool onto an unexamined process and hoped. The models were never the bottleneck. The implementation was. Anthropic just made that official and attached a billion-dollar price tag to fixing it.
Why this is a small-business advantage
Here is the plot twist the enterprise press keeps missing. The thing Ode sells for a fortune, small businesses can largely do for themselves, because the reason it is so expensive is a set of problems you do not have.
A Fortune 500 needs “special forces” engineers because it is carrying decades of legacy systems, a security review for every change, five committees per decision, and 10,000 employees who all do the same task eleven different ways. Implementation there is archaeology. You run a business where the person who understands the process, approves the change, and has to live with the result is the same person: you. That is not a weakness. In this specific race, it is the single biggest advantage on the board.
Taylor said it himself: “Non-AI companies are going to be among the big winners of this whole AI moment if they adopt the technology the right way.” He was talking about corporations that can afford his engineers. But the logic runs harder in your favor. You can decide to rewire how you handle quotes this afternoon and have it running by Friday. No Blackstone required. If you are still deciding which tools to build on, our guide to how to start a business with AI is a good place to anchor.
The implementation playbook you can run for free
Strip away the private equity and Ode’s method is four moves. You can run all four this week.
1. Pick one process, not ten. Ode’s ideal engagement is a single priority the CEO cares about most, not a company-wide “AI transformation.” Do the same. Pick the one task that eats your week: the quotes you retype, the invoices you chase, the same six customer emails you answer forever. One process. Name it.
2. Rewire it, don’t decorate it. The failure mode is adding a chatbot next to a broken workflow. The win is redesigning the workflow around what AI is good at. If you spend three hours a week turning call notes into follow-ups, the fix is not “use AI to write faster.” It is a system where the notes become a drafted follow-up automatically, and you edit instead of create.
3. Measure the impact, honestly. Ode’s actual secret is that it runs constant evaluations to prove the AI moved a business number. Copy that discipline. Time the task before and after. If “the AI way” is not clearly faster or better in two weeks, kill it and try a different process. No sunk-cost loyalty.
4. Keep a human on the last step. Every serious implementation, Ode’s included, keeps a person between the model and anything that matters. Let AI draft the proposal, the reply, the first pass. You approve what goes out the door. That one rule captures most of the speed and almost none of the embarrassment. The same principle powers the new wave of AI agents everyone is suddenly talking about.
Where the risk actually is
Taylor described AI as a “magic, hallucinating ingredient,” which is the most honest three-word summary anyone at a billion-dollar AI venture has offered. Magic when the process around it is sound. A liability when it is not.
The real risk for a small business is not falling behind on models. It is quietly rewiring a core process around a tool that is confidently wrong ten percent of the time and not noticing until a customer does. Before you hand any workflow to AI, it is worth pressure-testing which of your tools and processes are actually exposed. Our SaaS AI risk audit walks through exactly that. Implementation is leverage in both directions. Ode charges what it charges because doing this badly is expensive, and that is true at your scale too.
The bottom line
Siegel offered one more line that lands differently for a small operator than a corporate one: “It has never been an easier time to become an entrepreneur.” The billion-dollar version of that sentence is a joint venture with 100 elite engineers. The small-business version is you, one process, and a free afternoon. The labs just told you where the value is. You are closer to it than the Fortune 500 is.
Frequently asked questions
What is Ode with Anthropic?
Ode is a roughly $1.5 billion AI implementation company launched in May 2026 as a joint venture between Anthropic, Blackstone, Goldman Sachs, and Hellman & Friedman. It is built around the acquired startup Fractional AI and helps large companies deploy AI inside their real operations rather than just buy access to a model.
Does “implementation beats models” mean the model I choose doesn’t matter?
It matters less than most people assume. As Ode’s chief technologist put it, model choice is like picking a programming language: a real decision, but not where most of the value is created. For a small business, the leading models are close enough that how you wire one into your workflow matters far more than which one you pick.
How do I start implementing AI without a budget?
Pick a single repetitive process, redesign it so AI does the first pass and you approve the final step, then measure whether it actually saved time within two weeks. If it did, keep it and move to the next process. If it did not, drop it. You do not need consultants to run that loop.
Related Coverage
- How to Start a Business with AI — the tool foundation to implement on top of.
- SaaS Tools AI Risk Audit — spot the processes that are exposed before you rewire them.
- AI Agent Showdown — the tools doing the actual implementation work.
Faceted Media Magazine covers business, AI, and entrepreneurship for the people building what’s next.