AI readiness for owner-led companies
You bought the tools. The work did not change.
Most AI projects start with a tool. We start with the work: the outcome, the decisions, the handoffs, the exceptions, and the evidence. Then we automate only what should be automated, and say so plainly when the answer is not AI.
Free. No deck.
- Evidence-cited, not vibe-based
- Bounded scope with stop conditions
- Human authority on consequential decisions
- The answer is often not AI
You are probably here because someone said one of these
- “We bought AI tools but work has not changed.”
- “This process depends on one person.”
- “Growth is creating rework or handoff failure.”
- “We cannot agree which system or number is correct.”
- “We want an agent to do this entire process.”
- “We need to scale without adding the same headcount.”
- “An audit, customer, or incident exposed a workflow gap.”
The method
Clarify, simplify, standardize, then automate what is left.
In that order, deliberately. Automating a process nobody agreed on just makes the disagreement faster.
- 01
Name the outcome, not the tool
One workflow, one sponsor, one measurable consequence. If we cannot name what would be different, there is nothing to build.
- 02
Collect normal cases and exceptions
Standardized intake from the people who actually do the work. The exceptions are where the time goes, and they are the part every tool demo leaves out.
- 03
Decide what each step should be
Human, deterministic software, AI-assisted, or someone else's system. Most steps come back deterministic, and that is a good result.
- 04
Design the controls before the automation
What evidence is required before anything acts, who approves, what happens when it is wrong, and how you would know.
- 05
Hand over a decision, not a proposal
A ranked recommendation with its economics, its risks, and explicit stop conditions, including the recommendation to do nothing.
What you actually get
A line-by-line answer to “which of these steps should a machine be doing?”
This is the artifact the engagement produces, worked through one step at a time with the people who do the job. An illustrative example:
| Step | Owner | Why |
|---|---|---|
| Request arrives by email | Deterministic | Parsing and routing are rules. A model here adds cost and a failure mode, not capability. |
| Classify intent and urgency | AI-assisted | Language judgment on unstructured text. The model proposes; the queue it lands in is checkable. |
| Check entitlement and history | External system | The system that owns the record answers the question. Copying that state creates a second truth. |
| Approve an exception | Human | Consequential and contested. Authority stays with a person who can be asked why. |
| Draft the response | AI-assisted | Drafting is where language models actually earn their keep. Under review, never sent unread. |
Reasonable objections
The questions worth asking first.
- Can you just build us an AI agent?
- Possibly, but not responsibly before the workflow, its exceptions, its evidence, and its controls are clear. The engagement determines the smallest useful intervention, and it is often simpler than an agent.
- Why pay for diagnosis instead of implementation?
- Because unclear workflows transfer ambiguity into expensive implementation. You get an owned decision package and a bounded path, even when the recommendation is to simplify or to stop.
- Are you an AI agency?
- No. AI is one intervention class alongside elimination, standardization, conventional automation, approval-bounded execution, and leaving the work with a person.
- Can you promise a return?
- No. We define baselines, assumptions, acceptance criteria, and controls, then measure. Outcomes that depend on your data, systems, and people are not ours to promise.
- What does it cost?
- Engagements are fixed-fee and bounded, and the number depends on the workflow. The scoping call is free and ends with a straight answer about whether this is worth doing at all.
Start with one workflow.
Thirty minutes, no deck. Bring a process that costs more than it should and we will tell you honestly whether there is something here worth paying for.