The fastest AI wins were already paid for.

The situation

A leading London law firm had done everything a firm is told to do. The platforms were bought: relationship intelligence, email marketing, content distribution, Copilot on the desks. The team had capable people, some already building their own AI agents. The goals were clear. And still, very little was converting into changed work.

What we found

We built the evidence properly: a twelve-person workshop, six discovery deep-dives, partner and IT interviews, and structured challenge sessions with three of the firm's vendors. Over 300 candidate solutions were assessed and 67 ranked.

The finding fit in a sentence: the layer that converts investment into working AI, named ownership of each tool, documented process, a cadence with each vendor, was nobody's job. One content platform was costing £60,000 a year against a handful of posts a month, renewing on autopilot, with an AI module already included in the contract that had never been switched on. The firm's own people had built six AI agents that couldn't be shared, because nobody governed them.

What changed

The advice was to activate, in a specific order. The content platform kept its place, with a named owner, a single-channel strategy, and the AI module switched on. The home-built agents were brought under governance and rolled out across the team instead of living on one licence. Relationship intelligence was routed into the tools partners already work in, with no new logins and no behaviour change asked of them.

Most of the top ten moves needed no new spend at all. The firm's twelve-month AI roadmap now runs on people it already employs and contracts it already pays for, and every recommendation traces back to evidence the firm can audit.

Before you buy more AI, switch on the AI you own.