the method

Charted

My method for putting AI into a business without breaking it. Map the work, gate the risky parts, and keep checking the output after it ships. The discipline comes from lean process improvement. I have been applying it for twelve years, and it turns out AI is exactly the kind of unreliable process it was built to manage.

why a method

Most AI projects fail, and it is almost never because the AI was not smart enough. They fail because someone automated a process nobody understood, skipped measuring how the work went before, and had no way to check the output after. AI does not fix a messy process. It runs the mess faster.

So my method spends its energy on the process, not the model. The short version: map it, gate it, check it.

The six phases

  1. Discover

    1 wk

    Short interviews with you and the people who actually do the work. Where does the time really go, and which parts does everyone quietly hate?

    you get

    A written picture of how the work moves today, in your words.

  2. Map

    1 wk

    One working session. Your core process goes up on the wall: every step, handoff, wait, and workaround. We also mark what data each step needs and whether you actually have it.

    you get

    Your process, mapped and annotated. Most owners have never seen their business on one page before.

  3. Prioritize

    3-4 days

    Every automation idea gets scored on payoff, feasibility, data, and risk. Some steps get eliminated instead of automated, because automating a bad step just makes the bad thing happen faster.

    you get

    A scored shortlist and a 90-day roadmap with one recommended pilot.

  4. Pilot

    4-8 wks

    I build the winner with one success metric and guardrails attached. Anything risky, like sending money, contacting customers, or deleting records, waits for your approval before it runs. When the AI hits something it does not recognize, it stops and brings in a person.

    you get

    A working system measured against how the work went before it existed. We agree up front what scale, fix, or stop looks like.

  5. Scale

    1-2 wks

    Your team gets trained on the system, the runbook gets written, and ownership transfers to you.

    you get

    A system your business runs without me.

  6. Assure

    monthly

    The part most AI consultants skip. Every month I pull a sample of what the AI actually produced, score it against the standard we set in the pilot, and send you a one-page report. AI output drifts over time as your business and the models change. Checking is the only honest answer to that.

    you get

    A monthly report with real numbers, and drift caught before your customers feel it.

Phases 1 to 3 are the audit. Phase 4 is the build sprint. Phase 5 closes either one. Phase 6 is the assurance retainer, and it's optional: the system is yours either way.

What “assured” means

Three promises, built into every system I ship. They are not policies. They are how the software is constructed.

gated

Nothing risky runs without your approval

The systems I build are designed so they cannot take a dangerous action on their own. In my own products, billing fires only after success, so an agent has no path to charging a customer for work that failed. Your build gets the same kind of gates.

supervised

The AI knows when to stop

When a system hits something it does not recognize, it stops and hands the situation to a person instead of guessing. Factories have worked this way for a century. Your AI should too.

checked

The output gets checked, with numbers

I do not ask you to trust the AI. I sample its output every month and score it against a standard we agreed on. You see the score. If it slips, you know before your customers do.

what I won't do

  • I will not automate a step that should not exist. Elimination comes before automation, every time.
  • I will not quote you the scary industry statistics. Most of them have no source. Your numbers come from your own baseline, measured before anything gets built.
  • I will not do big-bang launches. Every system runs in the background first, then on a small slice, then fully. Promotion is earned at each step.
  • I will not sell you AI where a simple rule does the job cheaper.

where this comes from

This is not a framework I read about. My own AI platform, Vora, runs at 65 small businesses today, with these gates and checks built in, because I needed them before I named them. And the mapping discipline comes from a decade in enterprise financial services and infrastructure, where I used the same approach to cut delivery times by 29% before AI was part of the toolkit.

See how I build

Phase 1 starts with the audit

Audits start at $750, fixed scope. Two to three weeks inside your business, and you end up with the map, the scored shortlist, and one recommended pilot, whether or not I build anything.

Let's talk