Floburn.
Floburn Inc — Harness Map Readiness

Where does your business actually sit on the Harness Map?

Twelve self-assessment prompts across the five stages of how a business actually uses AI. Two screens of reading. No score, no email capture, no follow-up sequence.

Answer each prompt honestly — *yes,* *no,* or *sort of.* If you're unsure, the answer is *no.* The stage at which you stop being able to answer *yes* is, conservatively, the stage your business is at.

←Back to the journal
  1. 01

    Browser-tab AI

    1. 01

      Do members of your team open ChatGPT, Claude, or another AI assistant in a browser tab during the workday, on their own initiative?

    2. 02

      Is there any record of which AI tools your people use, and what they use them for?

    3. 03

      Has anyone on the leadership team set policy on what is and isn't allowed, with respect to AI use at work?

    If most answers are yes: Your people are experimenting. The leverage is invisible to the business. The risk is also invisible: shadow tooling, no audit trail, and a real probability of data leaving the org in ways nobody can trace.

  2. 02

    Tool-assisted AI

    1. 01

      Are at least three specific AI-enabled SaaS tools paid for by the company and used by named employees?

    2. 02

      Have any of those tools generated a measurable time saving you can name in hours per week?

    3. 03

      Is IT or operations responsible for those tools' performance, integrations, and renewals — or do individual users own them in the shadows?

    If most answers are yes: You're getting individual time savings, but the business operating model hasn't changed. This is where most companies get comfortable and stay. The cliff to Stage 3 is wide, and almost nobody crosses it from the inside without dedicated help.

  3. 03

    Workflow AI

    1. 01

      Is there at least one named business workflow — intake, support response, research, scheduling, content production — that has AI inside it as a load-bearing step?

    2. 02

      Does the workflow produce outputs reviewed and acted on by humans, with a defined fallback when the AI fails?

    3. 03

      Has the workflow's success been measured against a baseline from before the AI was added?

    If most answers are yes: You've crossed the cliff. The work is genuinely different now. The next risk is rot — a workflow with AI in it requires monitoring, drift detection, and a named owner. Without those, Stage 3 collapses back to Stage 2 within a quarter.

  4. 04

    Agent AI

    1. 01

      Does your company run at least one named agent (the lead-research agent, the support-triage agent, the content-discovery agent) that operates on a schedule or trigger, not only when a human invokes it?

    2. 02

      Are the agent's outputs landing in systems of record — CRM, ticketing, payroll, ERP — and not just in inboxes?

    3. 03

      Is someone accountable for the agent's performance the same way they'd be for a junior employee?

    If most answers are yes: You're operating with named agents that do work without supervision. Most companies at this stage have invested in monitoring, ops maturity, and a culture of treating AI like staff. The next move is portfolio thinking — which agents to add, which to retire.

  5. 05

    Operating AI

    1. 01

      Is at least one of your business processes literally unrunnable without its AI component?

    2. 02

      Are new employees trained on AI workflows as part of onboarding, the same way they're trained on email or the CRM?

    3. 03

      Have you redesigned any process specifically because the AI was available — not added AI to an existing process, but redesigned around the AI's capabilities?

    If most answers are yes: The business runs on AI. Your operating model assumes its presence. The risk profile here is concentration — what happens when the model degrades, the vendor changes the API, or the cost structure shifts. The work is second-system planning and contractual resilience.

What to do next

The work depends on which stage stopped you.

  1. Stage 1

    Visibility before tooling. Until you can name what your people are using and on what data, new AI investment is risk, not leverage.

  2. Stage 2

    Pick one workflow to cross the cliff on. Don't pick the most exciting; pick the one with the cleanest baseline and the most patient stakeholder.

  3. Stage 3

    Name the owner before the next quarter. The biggest threat to a Stage 3 win is silent rot, and the only defense is a named human watching it.

  4. Stage 4

    Portfolio review. Which agents are working, which are theatre, which should be retired. Most Stage 4 orgs are running at least one agent nobody trusts anymore.

  5. Stage 5

    Second-system planning. Concentration risk is your dominant risk. Identify the three processes most exposed if your primary model goes down, and design fallbacks.

If the diagnosis is sharper than your last quarterly planning exercise, you're not alone.

Most companies show up between Stage 2 and Stage 3, looking up the cliff. The work of closing that gap is what Floburn does. Engagements run four to twelve weeks.

Book a discovery call→See what we ship→