AI Readiness Consulting: How We Built the Process Blueprint That Made a Building Supply Distributor Actually Ready for AI
- Ed Hitchcock
- May 19
- 6 min read
By Ed Hitchcock, Enterprise AI Systems Architect, SupplyTech Solutions
Most distributors we meet think they have an AI problem. They almost always have a process problem first. The leadership team has read the same Gartner deck and seen the same vendor demos, and they want copilots, agents, and automation. We walk the floor for two days, ask how a quote actually gets built, and find out three branches do it three different ways. None of them can tell us what it costs to produce that quote.
That gap is what AI readiness consulting actually solves. Not strategy slides. Not a maturity scorecard. A documented, costed, validated operating blueprint that AI can be trained against, automation can execute against, and an ERP can be configured against. Without it, every dollar spent on AI burns on top of an unstable foundation. We have watched that movie too many times.
This post walks through how we built that blueprint for a regional building supply distributor over a single engagement phase. It is the same work we run with every operator before we touch a copilot.
The Operator We Started With
The company is a multi-branch building supply distributor in the Southeast, roughly 180 employees across five locations, low nine-figure revenue, family owned, second generation in the chair. They sell to contractors, builders, and a growing DIY base. Their stack at engagement start: an on-premise ERP from the early 2010s, an Excel-driven quoting workflow, a CRM the sales team logged into once a quarter, and a SharePoint document library that nobody trusted because nobody could find anything in it.
The CEO had been told by two AI vendors that he was 90 days away from autonomous purchase order generation. He suspected that was nonsense. He was right.
What he actually had was an organization where the top ten recurring workflows lived inside the heads of six tenured employees, three of whom were within five years of retirement. He had no idea what any of those workflows cost to execute. He could not tell us how long it took to onboard a new inside sales rep, only that it felt like it took forever.
That is the real precondition AI readiness consulting addresses. Process tribal knowledge plus financial opacity plus retirement risk equals an organization that cannot safely automate anything yet.
What AI Readiness Actually Means in Practice
We use a 5-phase capability model with every distribution client. Phase 1 is documented and costed processes. Phases 2 through 5 (knowledge management, operational AI, modular cloud ERP, enterprise AI) cannot start until Phase 1 is real. That order is not a preference. It is a constraint.
The reason is mechanical. An AI agent that processes purchase orders needs a deterministic spec of how a purchase order is created, validated, escalated, and paid. A copilot that drafts a quote needs a labeled corpus of approved quotes with the rules that produced them. A workflow automation that routes returns needs a documented exception path. None of that exists when work is tribal.
So the readiness work is concrete. We document every repeatable workflow, define ownership, map handoffs, and quantify the labor, time, and cost required to execute each one. We do it on the floor, with the people who actually do the work, not in a conference room with managers who think they know how the work happens.
That is the whole job. It is unglamorous. It is also the highest-leverage thing a mid-market distributor can do before spending a dollar on AI.
How We Ran the Engagement
We structured the work in five concrete actions, run over roughly eight weeks for the building supply distributor. Each action produced a deliverable that became raw material for the later phases.
First, we ran Gemba walks across both operational and corporate workflows. Two of us, one full week per branch for the first two branches, two days for each of the remaining three. We watched the work, asked clarifying questions, and recorded the actual sequence of steps including the workarounds nobody talks about in a status meeting. The point of a Gemba walk is to see the process as it is, not as the org chart says it should be.
Second, we generated Lean value stream maps for the most critical workstreams. For this distributor the priority list was quote-to-order, purchase order processing, customer returns, special order management, and month-end inventory reconciliation. Each map showed cycle time, wait time, rework loops, and the points where data moved between systems by copy-paste.
Third, we built a costing model. Fixed and variable costs were assigned to every workflow using a combination of payroll data, time-tracking estimates from the Gemba walks, and allocation ratios for shared overhead. The model gave the CEO his first real number for what a quote costs to produce (roughly 38 dollars in fully loaded labor before margin, which surprised him), what a return costs to process (between 22 and 41 dollars depending on whether it required a restocking trip), and what month-end inventory reconciliation costs (eight people, four to six days, somewhere between 16,000 and 22,000 dollars in labor every month).
Fourth, we generated standardized templates for SOPs, workflows, and training material. One template per artifact type. Pre-populated metadata fields so the documents would be machine-readable when the knowledge management system came online in Phase 2. This is the boring step that most consultants skip and most clients regret later.
Fifth, we documented all processes and procedures granularly enough that an AI model could later parse them as structured data. Step number, actor, system, input, output, decision point, exception path. Roughly 140 workflows ended up in the library by the end of the engagement.
The Tools We Used and Why
We deliberately kept the toolset minimal. The point of Phase 1 is not to introduce new technology. It is to use the technology the client already has correctly.
For documentation we standardized on the Microsoft 365 Premium suite in online mode rather than desktop. The client had been licensing M365 for years but only the operations team had moved to the cloud version. Switching the whole company to synchronous online editing eliminated the version-control chaos that comes from emailing docx files around. Documents version themselves indefinitely in SharePoint, which means we get a real audit trail and the work becomes safe to edit collaboratively.
For reporting we used Power BI. The sales team already had one dashboard the CEO trusted. We built a second one that tracked process performance: cycle time, labor utilization, and cost per activity, refreshed automatically from the time-tracking data we gathered during the Gemba walks. That gave leadership a continuous view of the metrics we had previously captured as a one-time snapshot.
That was it. No new SaaS contracts. No procurement cycle. Two products the client already paid for, configured correctly for the first time.
What Changed When the Blueprint Was Done
Three things shifted measurably by the end of the engagement.
The first was visibility. The CEO went from having gut estimates to having numbers. Top ten workflows were costed, mapped, and owned. The board meeting that followed the engagement was the first one where the operations slide showed cost per activity instead of generic efficiency commentary.
The second was time recovery. Several workflows had duplicate steps or unnecessary approvals that nobody had questioned in years. Eliminating the obvious ones (one approval layer on quotes under a 5,000-dollar threshold, one duplicate data entry between the CRM and the ERP that had been done by hand for six years) recovered roughly 10 to 15 hours per week of administrative capacity across the team. Conservative estimate. The real number is probably higher but we only count what we can defend.
The third was onboarding readiness. The training templates plus the documented procedures cut the time-to-productivity for a new inside sales rep by an estimated 10 to 20 percent based on the two reps hired during the engagement. Not a controlled experiment, but the managers who ran the training said the difference was obvious.
The harder result is the one you cannot put a number on. The CEO now knows what to spend AI dollars on. The retirement risk is bounded because the tribal knowledge is on paper, in a structured format, owned by a department. The next two phases (knowledge management system, then operational AI copilots) have a foundation to sit on top of.
What This Means for Other Mid-Market Distributors
Most of the AI consulting we see in mid-market distribution skips this phase and starts with copilots. Every one of those engagements we have inherited from another firm has had the same problem. The copilot answers questions confidently using stale, contradictory, or duplicated source documents. The agent generates purchase orders that match an undocumented exception path nobody told it about. The operations team loses trust in the system within 90 days and the project gets quietly shelved.
AI readiness consulting is not a strategy deliverable. It is the unsexy operational work that has to exist before any AI investment pays back. For a mid-market distributor, this typically means six to ten weeks of process documentation, cost modeling, and template work before phase two begins. We will not start a knowledge management build without it. We will not stand up an operational AI agent without it. The math does not work if the foundation is wrong.
If you are running a distribution business and a vendor is selling you autonomous workflows on top of an undocumented operating model, ask them what your top ten processes cost to execute today. If they cannot answer, neither can their AI.
That is the whole point of getting ready.
About the author: Ed Hitchcock is the founder of SupplyTech Solutions, where he designs and deploys AI-enabled operating systems for mid-market distributors and private operators. Previously a program leader at Amazon and a certified Lean Six Sigma Black Belt.
Comments