Where to start
AI Enablement for Knowledge Work
We put the firm on the right platform, teach people to use it well, and show you in real numbers what it returns. Afterward, monthly advisory keeps you current as the tools change. Where a platform can’t do the job, custom builds are scoped separately.
Twenty-five minutes. Tell us where the firm is spending its time, and we’ll tell you honestly what AI can and can’t do about it, including when the answer is “not much, yet.”
How it runs: four phases, 4–6 weeks
-
Weeks 1–2Discover
-
Weeks 2–3Stand up
-
Weeks 3–5Enable
-
Weeks 5–6Measure
Weeks 1–2
Discover
We interview your people and watch the real work, not a demo of it. We measure before we change anything.
Typical starting points: pulling numbers from PDFs into tax forms, finding context buried in email, or drafting proposals from your own templates.
You get: An opportunity list and a measured baseline on selected workflows.
Weeks 2–3
Stand up
Claude, ChatGPT, Gemini or Copilot, chosen on the merits. We design the configuration; your IT or MSP executes it. Clear rules for what can go into AI, the right model for each kind of work, and costs that stay predictable.
You get: An enterprise AI platform, configured by your IT to our design and owned by you. Data-handling rules, which model to use for which work, and the spend controls your platform offers.
Weeks 3–5
Enable
Your templates, methodology and technical positions, made retrievable. Role-based workshops on prompting well: better first drafts, fewer reruns, less wasted usage. A working prompt library and an internal champion.
You get: An assistant that answers from your firm’s own material. Role-based training, a working prompt library, and a trained internal champion.
Weeks 5–6
Measure
We take a first read at handover and set up a monthly show-back: hours back, set against what the firm spends on AI. Real numbers, not vendor claims.
You get: A first read on return, a monthly show-back, a costed 12-month plan, and an executive readout.
Who does what
- AI Optimized Group
- Chooses the platform with you and designs the configuration, guardrails and spend controls.
- Your IT or MSP
- Executes the configuration to our design.
- Your firm
- Owns the platform. Subscriptions are bought in your name.
What we need from you
- A partner who owns it: one partner as sponsor, from kickoff to the readout.
- People and access: the people we’ll interview and the workflows we’ll observe, named in week 1.
- Your IT lead or MSP: to execute the configuration to our design during Stand up, weeks 2–3.
- Subscriptions: bought in your name once the platform is chosen.
- An internal champion: someone who takes over at handover.
What’s included, and what isn’t
Fee
Fixed fee. No hourly billing.
- Included
- Workshops, configuration design, knowledge layer, guardrails, measurement and roadmap.
- Subscriptions
- Bought in your name. We take no margin on what you buy.
- Not included
- This is not a software project. Custom builds are scoped separately.
- Afterward
- Monthly advisory: each month, what’s sticking, what’s noise and what’s worth piloting next, with flexible hours for calls or ad-hoc evaluations. Priced separately.
Starting smaller
A one-week paid discovery gives you the opportunity list and our recommendation, credited if you proceed.
Questions partners ask
We already have Microsoft 365. Isn’t the answer just Copilot?
Not necessarily. Claude, ChatGPT, Gemini or Copilot, chosen on the merits; being on a suite doesn’t decide it. In one recent engagement, Microsoft’s own tools meant more cost and complexity for what the firm needed, and we recommended something else.
Who owns the platform?
You do. It’s configured by your IT to our design, and subscriptions are bought in your name. We take no margin on what you buy.
What happens to our client data?
It stays under your control. The platform is a business account bought in your name and configured by your IT, not anyone’s personal account. In Stand up we set clear rules for what can go into it and what stays out, starting with the firm’s own material before any client matters. We choose a plan whose terms keep the firm’s data out of model training. It’s designed in from the start by security experts, CISSP-certified, not added later.
Our IT provider already offers AI training. How is this different?
Keep them. They execute the configuration to our design. What we add: the platform chosen on the merits, your firm’s own material made usable, and a measured read on what it returns. We don’t sell platforms or seats, so the recommendation doesn’t depend on anyone’s catalog.
How do you measure return?
On your own workflows. In Discover we time a few of them: hours per week and turnaround. At handover we time them again, then report monthly. If a workflow didn’t move, we say so.
Do our people need to be technical?
No. Training is by role, built on the work each role already does. The internal champion needs time set aside for it, not a technical background.
Do you come on site?
We work remotely by default, and the fixed fee assumes it: interviews and observation happen over video, and workshops run live online. If you’d like us in person for some of it, we can arrange that.
What does it cost?
A fixed fee, with no hourly billing. We set out the figure in the first conversation, along with the one-week discovery.
What if AI can’t do much for us yet?
Then we’ll tell you. “Not much, yet” is a possible answer, and we’d rather give it in twenty-five minutes than after six weeks.
Do you build software?
Not in this engagement. This is not a software project. Custom builds are scoped separately.
What happens after week 6?
You have a costed 12-month plan and the monthly show-back we set up. Monthly advisory: each month, what’s sticking, what’s noise and what’s worth piloting next, with flexible hours for calls or ad-hoc evaluations. Priced separately.
Self-assessment: Before you buy AI seats for the firm
Start with a conversation
Twenty-five minutes. Tell us where the firm is spending its time, and we’ll tell you honestly what AI can and can’t do about it, including when the answer is “not much, yet.”
Peter Barca, Co-founder