Tools are a commodity. Orchestration is the edge.

• AI moved from novelty to leverage in under two years.
• The firms pulling ahead aren't the ones with the most tools — they're the ones whose people know how to orchestrate them.

Three risks, named plainly.

1. Capability gap — smaller, faster teams out-execute you.
2. Talent risk — your best people leave for firms that give them AI leverage.
3. Governance trap — ungoverned AI use creates confidentiality and accuracy exposure.

What is an hour of your time worth?

• Put a real number on a senior hour — you already do. It's the charge-out rate.
• Any task worth less than that rate is a task to hand to a system.

This reframes delegation as economics, not enthusiasm.

Don't bet the firm on one model.

• Keep two or three capable models, not one.
• Resilient to any single vendor's outages, price moves, or hallucinations.

Teach the roles, not the brands.

• Reasoning / drafting — a frontier chat model.
• Research / sourcing — a model with live, cited web access.
• Agentic execution — a model that can run multi-step tasks.
• Tool names change quarterly; the roles don't.

The difference between "the AI said so" and a reviewed, defensible result — the single most valuable habit for a risk-averse firm.

At least one agent, for everyone.

• Every team member gets an agent handling the non-billable admin.
• Inbox tax, research, meeting prep, follow-up — the leak, automated.
• Frees seniors for the judgement work only they can do.

What to delegate first.

• Start with the highest-frequency, lowest-judgement tasks.
• Triaging email, summarising documents, drafting routine correspondence.
• Keep a human on anything carrying legal or client-relationship weight.

Onboarding, precedent, house style — searchable, reusable, owned by the firm. Where your IP compounds instead of walking out the door.

Reviewed and ready — or it's a zero.

• Work is either reviewed, approved, and ready — or it doesn't count.
• "I'll finish it later" is how billable weeks leak.

Draft, review, ship in one sitting.

1. Capture the thinking once — voice or transcript.
2. Let AI produce the rough first version immediately.
3. A human refines with judgement.

A client letter, a proposal, a marketing asset — minutes, not days.

Show the human-in-the-loop edit, not a magic-button claim. Demonstrated, not promised.

You're not training a public model.

• Enterprise / sandboxed plans only.
• Confirm the data policy before loading anything client-sensitive.

Answers the first objection every professional-services buyer raises.

• Cross-Model Verification on anything that leaves the building.
• A human review gate on legal and client-facing output.
• An audit trail of what the AI produced and who approved it.

Governance is the feature, not the friction, for this buyer.

Partner-to-partner, confidential, no hype. The relationship contract for the sprint.

1. Hidden money — pricing, dormant assets, uncollected revenue accessible this week.
2. New revenue — an offer or strategy already in front of you, unexploited.
3. Partner revenue — JV, channel, referral.

Probe all three; run an agent live on one.

In a capability sprint we put a real number on it, redesign the workflow with AI doing the heavy lifting, and leave the team able to run it themselves.

A quantified roadmap.

• Leave every engagement with a roadmap — whether they buy the next step or not.
• Value demonstrated, not promised.

1. AI Discovery Conversation — paid, 60–90 min. Find the money, the risks, a roadmap.
2. 1-Day Orchestrator Workshop — your team, fully enabled.
3. 3-Day Capability Sprint — find the money and build the systems.

• A personal stack and at least one agent per team member.
• A shared second brain of institutional knowledge.
• A quantified revenue roadmap + the built systems to capture it.
• The capability to run it again — without us.