The billable hour just hired its own undertaker.
The Summary
- OpenAI made Telon, a six-month-old London startup, a "select partner" to embed "legal engineers" inside law firms — mostly ex-lawyers who configure models, write prompts, and train attorneys to use AI
- The real insight: selling AI to lawyers isn't the hard part anymore. Getting them to use it is. So OpenAI is deploying humans to solve the human problem.
- Telon's model flips traditional consulting: instead of multi-month McKinsey engagements to install software, you get forward-deployed engineers who live inside your firm and make the tech work daily
The Signal
OpenAI wants lawyers to use its models. Not just buy them — *use* them. So it's doing something that sounds almost quaint in the AI era: sending actual people into law firms to show attorneys how.
Telon is less than six months old, but it already landed a spot in OpenAI's Partner Network, which launched in June with a stated goal to certify 300,000 consultants by the end of 2026. The startup's pitch is simple: hire us, we embed former lawyers inside your firm as "legal engineers," and they make your AI stack actually productive. Think of it as forward-deployed engineering-as-a-service, but for the industry that invented the six-minute billing increment.
"The hardest part about selling AI to law firms isn't closing the deal. It's getting lawyers to actually use it."
Lewis Bretts, Telon's founder and a former trial attorney who built PwC's legal practice, saw this friction firsthand. Law firms are buying AI tools. Corporate legal departments are signing contracts. But between purchase and productivity sits a chasm: lawyers don't know how to write good prompts, configure agents, or trust outputs enough to rely on them for client work. The tech sits idle. The ROI never materializes. The partners get skeptical.
This is where Telon's model diverges from traditional professional services. The old playbook: hire McKinsey or Deloitte, get a multi-month engagement to roll out Salesforce or another enterprise platform, then wave goodbye as the consultants leave and your team muddles through. The new playbook:
- Embed an engineer who used to practice law inside your firm
- They configure OpenAI models for your specific workflows
- They train your attorneys on prompt engineering and agent design
- They stick around to iterate as cases evolve
Bretts said he started Telon after noticing enterprises were shifting away from those big-bang implementations. The work is changing. It's less about installing a database and more about teaching knowledge workers to co-pilot with agents. That requires someone who speaks both languages: legal reasoning and model configuration.
OpenAI's partner strategy reveals something critical about the agent economy. Building powerful models is table stakes now. The real competitive moat is adoption velocity. If your customers buy your software but don't use it, you lose to whoever can actually get attorneys billing hours with AI assistance. Telon gives OpenAI human infrastructure to solve the last-mile problem: turning skeptical lawyers into daily users.
The Implication
Watch for this pattern to spread beyond legal. Every knowledge-work vertical has the same adoption gap: people buy AI tools, then struggle to integrate them into actual workflows. The companies that win won't just ship better models. They'll ship people who know how to make those models productive in context.
For lawyers specifically, this changes the game. If you're a mid-tier associate doing document review or contract drafting, your firm is about to embed someone whose job is making AI better at your job. The billable hour model starts cracking when agents can do research in minutes instead of days. Telon's engineers aren't there to protect associates. They're there to make the tools work well enough that firms can do more with fewer people.
If you're building in the agent space, note the wedge: don't just sell software. Sell outcomes. And if your customers need help getting to those outcomes, build or partner for the human layer that closes the gap.