What Ogletree Deakins Learned the Hard Way About Rolling Out AI

Most AI adoption content is either a vendor pitch or a vague success story. Tim Fox's session was neither. It was an honest account from inside Ogletree Deakins of what failed first, what actually worked, and a clean decision framework for buying versus building that any professional services firm could use tomorrow.

What Did Not Work: Four Ways AI Rollouts Fail

Fox opened with the failures, not the wins, which is already rarer than it should be at a conference. Four specific ways Ogletree Deakins' early attempts fell apart:

1. Build it and they will come. Availability is not adoption. Making a tool accessible doesn't mean anyone uses it.
2. Understaffing. The assumption that you hire people first and adoption follows on its own doesn't hold. Adoption needs active support, not just headcount.
3. One audience. A law firm, as Fox put it, is at least two organizations. What works for attorneys doesn't automatically work for staff, and treating the firm as one homogeneous audience means missing both.
4. Analysis paralysis. There's no risk-free option. Waiting for a perfectly safe path to reveal itself just means never launching.

His line summarizing the whole lesson: "The launch is the beginning of the deployment, not the finish line." Getting a tool live is step one of a much longer process, not the end of the project.

What Worked: A Real Adoption Playbook

Once Ogletree Deakins corrected course, four things actually moved adoption forward:

1. Peers teach peers. Not top-down training sessions, colleagues showing colleagues what they'd figured out.
2. Measure and report. Success begets success. Visible wins get shared and referenced, which builds momentum on their own.
3. Change management. Framed explicitly as two-thirds people, one-third technology. The harder part of a rollout is never the software.
4. Shadow AI. Buy what people are already using. If staff have already found a tool that works for them informally, that's a signal, not a problem to shut down.

The data Fox shared behind why peer influence matters so much: people were 2.5x more likely to adopt AI when a leader used it, 3x more likely when a teammate did, and 5.5x more likely when a peer from a different part of the organization did. That last number is the one worth sitting with — cross-functional peer influence beat both leadership example and direct teammate influence.

Buy Versus Build: A Framework Worth Stealing

This is the most immediately useful part of the whole session, and it translates cleanly outside of law firms. Fox's rule for deciding what to buy versus build:

Buy when: you need frontier capability, a domain-specific platform already exists, or a niche tool already does the job well. His own examples: Claude, Copilot, Harvey, Lexis Protégé.

Build when: the workflow is genuinely unique to your organization, the volume justifies the engineering investment, or no vendor serves that specific need. His examples: internal automations, internal ops, immigration-specific workflows.

The rule in one line: "Buy where we can. Build where we cannot."

Fox didn't stop at the framework, he added the practical caveats that make it usable:

• On building: it's never one and done. Someone has to support it on a Friday afternoon when it breaks. A weekend is enough to build something, but months are what it actually takes to deploy it properly.
• On buying: price in the switching cost before you sign anything, and keep contracts short. Locking into a long-term vendor agreement before you know if the tool sticks is a common, avoidable mistake.

What to Take Back: Five Concrete Actions

Fox closed with a direct list, meant to be actionable immediately:

1. Staff adoption as its own function, not an afterthought bolted onto IT.
2. Let peers teach peers instead of relying only on formal training.
3. Measure and champion your usage so wins are visible and referenceable.
4. Buy where you can, build only where you genuinely have to.
5. Build governance before adoption, not after something goes wrong.

Why This Framework Travels Beyond Law Firms

None of this is law-firm-specific advice dressed up in legal language. Every professional services organization, agencies included, has the same structural problem: multiple internal audiences with different needs, real switching costs on vendor decisions, and adoption that lives or dies on peer influence rather than mandates. The buy-versus-build framework alone is worth adopting wholesale, and the adoption multipliers Fox shared (especially that cross-functional peer number) should change how any rollout plan gets structured.

Rolling out an AI tool and not sure if adoption will actually stick? Winsome helps companies build the change management plan that makes the difference between a tool people use and a tool people ignore. Talk to Winsome about your AI rollout.