far point ai
Most firms buy the software. Then they watch it die.
Not because the tools are weak. Because nobody changed how the team works. The licenses go live, the kickoff email goes out, and three months later two people are using it, and one of them signed the invoice.
The method is the lever. We install AI inside professional services teams of roughly 5 to 50: one person's win at a time, a human reviewing everything before it goes out, and a scope you approve phase by phase. You're not buying software. You're buying the adoption.
the engagement
Three phases. Each one your call.
Each phase is fixed scope. You approve each one before it starts; nothing renews on its own.
- Foundation. Licenses live, the usage policy written and signed, the team roster confirmed. Light touch, and quick.
- Adoption. The core of the engagement. We find each person's win, then build the habit around it: thirty minutes a day, one experiment a week, for eight weeks. Busy season gets a blackout; if a heavy stretch hits, the clock stops and nothing slips.
- Production workflows. The wins your team proved by hand become workflows that run on their own, with a person reviewing before anything goes out.
Stop at any phase and keep everything built. The phase gates are your exit, open the whole way through.
how we run it
A gift, not a mandate.
Every failed rollout dies the same way: a decree from above. A mandate creates fear, and fearful people find quiet ways to never use the thing. A tool that makes your day lighter creates pull. So the first team session says the true thing out loud: there are parts of your job you don't enjoy, and this exists to take them off your plate.
Then we find each person's win. One task per person, something they already dislike, that AI takes over in the first few weeks. The person who spends Monday copy-pasting client details into outreach templates walks in to find the drafts already written and waiting. They review and send. Hours back in a week, from one change. That's the moment a team stops complying and starts asking "could AI do this part" on its own.
And a human stays in the loop. Nothing reaches a client without a person reviewing it first. Trust builds because nobody is asked to hand their name to a machine.
compliance
The rules come first.
Your firm holds other people's sensitive information, so the rules get written before the first experiment runs, not after the first mistake.
One hard rule from day one: no client data goes into any tool without the right agreements in place. That single rule removes most of the risk before we automate anything, and most early wins never need to touch sensitive data at all.
The money stays clean too. You pay all software directly, at the vendor's price. Nothing is marked up, and nothing hides inside our fee.
proof
We run our own business on this.
Far Point Strategy runs its own business on the Far Point Operating System built on Claude: marketing, client delivery, operations, the books. We're not reselling a method we read about. We're installing the one we run every day.
Every member of the first Edge Accelerator cohort, including Alysse Bryson, Tamar Routley, and Billy Gwaltney, runs their own Claude-based assistant built in the program. The same rollouts now run live in our free workshops, Thursday July 16, 23, and 30. You can watch the system work before you spend a dollar.
return math
Every projection travels with its inputs.
Every AI vendor will hand you an ROI slide this year. Ask to see the inputs. A number without its inputs is a guess wearing a suit.
Our projections travel with their formula and their named assumptions: hours recovered per person per week, valued at your loaded hourly rate, over the weeks the sprint actually covers. Where a figure is a guess, it's marked as a guess, and it gets replaced with your team's real data the moment the workflow runs. You can argue with every input. That's the point of printing them.
where you fit
Not a team of 5 to 50?
This page is for teams. If that's not where you are, the system still fits; it comes through a different door.
- Going solo. The Edge Accelerator installs the same system for one person: a 60-day live cohort where you build one service, one price, and your own Claude-based assistant, working alongside Preston Peterman. Cohorts are capped at eight. The next one starts Tuesday, August 4; applications close Friday, July 31.
- Running a firm. The Collective is the ongoing room: a standing cohort of owners working the same systems month over month. The details live on the Collective page.
next step
Start with a working session.
No proposal first. A working session: your team, one real workflow, built live, so you judge the method on your own work before any scope gets written. If it's not a fit, you keep the hour's build anyway.
Questions before you book? Write to team@farpointstrategy.com.
faq
Questions, answered
- What does an AI implementation engagement include?
- Three fixed-scope phases: Foundation (licenses, a signed usage policy, the team roster), Adoption (per-person wins built into a thirty-minutes-a-day habit over eight weeks), and Production Workflows (wins your team proved by hand turned into workflows that run on their own with human review). You approve each phase before it starts, and you can stop at any phase and keep everything built.
- What tools do you use?
- Claude is the core of the stack; we run our own business on the Far Point Operating System built on Claude, and we connect it to the tools your firm already uses. All software is billed to you directly, at the vendor's price, with no markup.
- How do you handle sensitive data?
- Rules first: no client data goes into any tool without the right agreements in place, and that policy is written and signed in the Foundation phase before any experiment runs. Most early wins never touch sensitive data at all.
- How long does adoption take?
- The adoption sprint runs eight weeks, at thirty minutes a day and one experiment a week per person. Busy season gets a blackout; if a heavy stretch hits, the clock stops and nothing slips.
- Do we need to be technical?
- No. The work happens in plain English at thirty minutes a day, and we build the technical plumbing; if your team can write an email, they can do this. A person reviews everything before it goes out, so nobody has to trust a system they can't see.