AI Readiness Report

Mound City Industrial Supply (sample)

01Executive Summary

Mound City Industrial Supply runs $7.4M through sixteen people, and the constraint on your next move isn't the warehouse — it's the back office. You told me the goal plainly: get to $10 million without adding office headcount, and get this business running without you in it sixty hours a week within eight years. Both of those goals hit the same wall, which is that too much of how this place works lives in your head and Rhonda's, and too much skilled time gets spent retyping paper your customers already wrote.

The money and time are leaking in four places you named with numbers: Deb and Tony re-keying 45-plus phone/email/fax orders a day into Prophet 21 at five-plus minutes each, with a couple of $150–200 wrong shipments a week on top; quotes taking a day or two in a market where your bigger accounts told you the first decent number wins; a full owner-day every month building the margin report by hand in Excel; and $380,000 sitting past 60 days while collections happen only in Rhonda's spare minutes. Separately, you're sitting on 500-plus dormant accounts that already know you — and a 2019 blitz that proved re-contact works.

What's genuinely working is the part that's hardest to build: a crew that knows the product cold, answers the phone well, and holds twenty-year relationships. That's not a weakness to automate away — it's the reason the fixes below are worth doing, because every hour you free from retyping is an hour those people spend where they're irreplaceable. Your constraints are the right ones: P21 stays, a human signs off on everything customer-facing, one proven win at a time, and nothing that can't outlive its builder.

This report ranks the fixes by the cost you're carrying today and the ease of proving them out. The order recommendation isn't accidental: quoting is where the fastest revenue win lives, order entry is where the biggest recurring time drain lives, and reactivation is your most de-risked growth lever. Every one of them fits inside your rules — alongside P21, with Rhonda's signature on the way out the door.

02AI Readiness Score

41/ 100

0100

Readiness measures how much groundwork exists today — not whether AI is worth it for you. A low score doesn’t mean ‘not ready for AI’; it means the gap is large, and the gap is exactly what the opportunities below are worth.

Build the foundation0–39
Start with the enabling groundwork — the roadmap’s first phase matters more than any single build.
Start with one workflow40–59
Start with the top opportunity and prove it out — one delivered win builds the case for the next.
Build in parallel60–79
The groundwork holds — two or three opportunities can move at once.
Optimize and extend80–100
A strong base — the gains come from compounding and extending what already works.

Process Maturity: 9/20; Core order-to-cash runs on a real ERP (Prophet 21), but quoting, margin reporting, and collections are manual and ad hoc — done 'when there's time.'; 0 — processes live in people’s heads · 20 — documented, measured, and consistently followed
Data Readiness: 10/20; Six years of orders, 900 accounts, and lane/pricing history sit in P21 — strong raw material, but locked in exports and Excel rather than flowing anywhere.; 0 — data scattered across inboxes and spreadsheets · 20 — centralized, current, and queryable
Technology Foundation: 8/20; P21 is a capable backbone and stays, but everything around it is Excel and email, with no integration layer and no technical staff.; 0 — disconnected tools and manual re-entry · 20 — modern systems with APIs that talk to each other
Team Readiness: 7/20; World-class on product and relationships, not computer people by your own account; one credible owner in Rhonda, but she's already at capacity.; 0 — no AI exposure and little appetite · 20 — already experimenting and asking for more
Strategic Alignment: 7/20; Unusually clear goals — $10M without office headcount, a transition in eight years — and hard constraints that make scoping easy.; 0 — no owner and no goal for AI · 20 — a named owner and a clear place in the plan

03Current State Assessment

Here's an honest read of where Mound City sits today, dimension by dimension, grounded in what you told me.

Process maturity. Your order-to-cash spine exists and works — P21 handles inventory, orders, and receivables. But the high-value processes around it are manual and event-driven rather than systematic. Quoting takes a day or two because someone checks cost, checks availability, sometimes calls the supplier, then writes it up by hand. Collections happen when Rhonda has a spare minute, which you said is never. The margin report is rebuilt from scratch in Excel every month. These aren't broken processes — they're undocumented ones that depend on specific people remembering how, which is exactly the risk you named for a transition.

Data readiness. This is quietly your biggest asset. You have six years of order and pricing history in P21, 900 accounts with purchase history, and known lane/product patterns. The problem isn't that the data doesn't exist — it's that it only comes out as an export you hand-assemble. That's good news: the raw material for quoting assistance, margin automation, and reactivation targeting is already sitting in your system.

Technology foundation. P21 is a solid backbone and it stays — you were clear it took two years to get everyone comfortable and you won't repeat that. The gap is that nothing talks to P21; everything else is Excel and email, and there's no one technical on staff. That shapes every recommendation: things sit alongside P21 and feed off it, they don't replace it.

Team readiness. Your crew is great at the human part of this business and, in your words, not computer people — and you're not turning them into programmers. That's a design constraint I take seriously. The bright spot is that one counter employee already uses ChatGPT for thread specs and cross-references and is fast with it, which tells me the appetite is there when a tool earns its place. Rhonda — nineteen years, thorough, skeptical of everything new — is the right owner precisely because if it survives her, it works.

Strategic alignment. You know where you're going and what you won't tolerate. Scale revenue without scaling office headcount; get the know-how out of two heads before you step back. The underlying pattern across all five dimensions is the same: strong human and data foundations, weak connective tissue between them. Every fix below is connective tissue.

04Priority Opportunities

1. Quote turnaround assistant — first decent number back, faster — the quick win.
The problem
You get 25 to 30 quote requests a week and each takes a day or two because someone has to check cost, check availability, sometimes call the supplier, then write it up. Your bigger accounts told you straight: the first decent number back usually gets the order. And you don't track win rate, so you can feel the losses but can't count them.
What it costs
You didn't price this one directly, so here's a conservative, adjustable build. At 27 quote requests a week, that's about 1,400 a year. You said the first decent number usually wins — assume that a faster turn recovers just 2 additional orders a week that today go to whoever answered first. At an average order value well below your blended figure — call it $600 — and a conservative 25% gross margin, that's $150 of recovered margin per order, times 2 a week, times 50 weeks = $15,000 a year in recovered margin. That is deliberately low; if faster quoting wins even one order a day, the figure triples. Swap in your own average quote value and close rate once you start tracking them — which is itself the first thing this fix gives you.
Estimated annual cost of the problem: $15,000.
Approaches
A P21-fed quote drafter with a one-screen approval
Build an assistant that turns a quote request into a ready-to-send draft in minutes. The trigger is a request landing by email or phone; the inside salesperson pastes or forwards the requested items into a simple intake screen. The assistant matches each requested part against your P21 catalog (handling the customer-part-number mismatches that slow you down today by learning cross-references over time), pulls current cost and on-hand availability from P21 via its data export or API, applies your pricing rules — contract price if the account has one, otherwise a margin target you set by product category — and flags any line where availability is short or cost moved recently so a human knows to call the supplier. The output is a formatted quote drafted against your six years of lane and pricing history, presented on one screen for Deb, Tony, or Rhonda to review, adjust, and send. Nothing goes to a customer without that signature — your hard rule, honored by design. Crucially, every quote is logged with its outcome, so within a quarter you finally have a win rate.
Contract-price guardrails so repeat quotes are near-instant
For accounts with negotiated contract pricing, the assistant should recognize the account and apply the agreed price automatically, turning a repeat spot-buy quote into a same-minute confirmation rather than a day-long lookup. This also catches the quiet margin leak of quoting an old price when a supplier cost has risen — the tool surfaces the delta before you commit.
The Meridian bridge
The quote-drafting logic, the P21 data connection, the customer-part-number cross-reference matching, and the one-screen approval flow are what we'd build and configure for you against your actual catalog and pricing rules. The document-generation and email-send pieces can lean on off-the-shelf components we'd select and wire in rather than build from scratch. We set it up, train Deb, Tony, and Rhonda on the approval screen, and hand it over running — no six-figure project, and it sits alongside P21, never touching your ERP setup.
The Meridian bridge is our answer to ‘so what do we do about it?’ — the specific thing we’d build for you, where the off-the-shelf part ends, and where the custom work begins.
2. Order-entry capture — stop re-keying documents your customers already wrote — the medium-term.
The problem
More than half of your 80 daily orders arrive by phone, email, or fax as customer POs, and Deb and Tony re-key every one into P21 — five or six minutes each when clean, longer when part numbers don't match yours. Mis-keys ship the wrong item about twice a week at $150 to $200 each.
What it costs
Two stacked costs. Time: roughly 45 manual orders a day at 5 minutes each is about 3.75 hours daily, call it 3.5 hours to be conservative — about 875 hours a year of skilled inside-sales time spent retyping. Valuing that at a $22/hour loaded cost, that's about $19,250 a year in labor. Errors: 2 wrong shipments a week at $175 each, times 50 weeks, is another $17,500 a year. Combined, this problem costs roughly $36,750 a year — and it's the one standing directly between you and $10M without adding office headcount. Adjust the labor rate to your real fully-loaded cost and the number moves with it.
Estimated annual cost of the problem: $36,750.
Approaches
Document-to-P21 extraction with a review-and-post screen
Build a capture step that reads incoming customer POs — emailed PDFs, scanned faxes, and email text — and extracts line items, quantities, customer part numbers, and PO number into a structured draft. The trigger is a PO arriving in a dedicated inbox (or the fax scan folder). An extraction model reads the document; a matching layer maps the customer's part numbers to your P21 SKUs using a cross-reference table that grows every time someone confirms a match, which directly attacks the 'longer when part numbers don't match' problem. The result lands on a review screen showing the original document beside the parsed order, with any unmatched or low-confidence line highlighted in a color that says 'look here.' Deb or Tony confirm and post to P21 with one action. This keeps your human-signature rule intact while cutting the five minutes of typing to under a minute of checking — and because the risky lines are flagged, it attacks the wrong-shipment rate at the same time.
Confidence-gated auto-posting for clean, known orders
Once the matching table has matured on your highest-volume repeat customers, let orders that parse at high confidence — every line matched, quantities sane, known account — flow to a lighter one-click confirm, while anything ambiguous still gets the full side-by-side review. This concentrates Deb and Tony's attention on the orders that actually need judgment rather than every order equally.
The Meridian bridge
The extraction pipeline, the growing part-number cross-reference table tuned to your customers' quirks, the confidence flagging, and the review-and-post-to-P21 screen are the custom build — this is the heart of the engagement and where our work lives. We'd start with a pilot on your two or three highest-volume PO formats to prove accuracy before widening. We build it, run it in parallel against manual entry until the numbers hold, then train your team and leave it in their hands.
3. Dormant-account reactivation — call the 500 accounts nobody calls — the quick win.
The problem
About 900 accounts have bought in the last five years and only around 350 are active in any given month, so 500-plus known accounts sit quiet because nobody calls them. You ran a phone blitz on dormant accounts in 2019, it flat-out worked, and you never kept it up because it was always the thing that could wait.
What it costs
You proved this works, which lets us estimate conservatively. Of 550 dormant accounts, assume a modest reactivation of just 10% — 55 accounts — brought back to even a small run rate of $2,000 in annual purchases each. That's $110,000 in reactivated revenue; at a conservative 25% gross margin, about $27,500 a year in recovered margin, recurring. This uses only accounts that already know you and a re-contact method you've already seen work — it's the most de-risked line in this report. Tune the reactivation rate and account value to what your 2019 blitz actually produced.
Estimated annual cost of the problem: $27,500.
Approaches
A P21-driven reactivation queue with drafted, personalized outreach
Build a recurring job that pulls from P21 every account with no order in 90+ days, ranks them by prior spend and recency so the best prospects surface first, and generates a call-and-email queue for whoever's working it. For each account the assistant drafts a personalized touch grounded in real history — what they used to buy, roughly how often, when they last ordered — so Deb, Tony, or a part-timer opens the day with a prioritized list and warm talking points instead of a cold spreadsheet. Every outreach and outcome logs back against the account, so you can see reactivation rate and stop guessing. Because your rule is a human signs off on anything customer-facing, the emails are drafts your person approves before sending, and the calls are, of course, human. This turns 'the thing that can wait' into a standing ten-account-a-day rhythm that doesn't depend on anyone having spare time.
Win-back sequencing that escalates only when it earns it
Structure the outreach as a light sequence rather than one shot: a first personalized email, a follow-up call a few days later for non-responders, and a final offer touch for the accounts that were once high-spend. The queue tracks where each account is in the sequence so no one falls through and no one gets pestered, mirroring what made 2019 work but without relying on memory to keep it going.
The Meridian bridge
The P21 pull, the ranking logic, the history-grounded draft generation, and the outcome logging are what we'd build. The email delivery and any call-list display can use off-the-shelf tools we'd select and integrate — you don't need anything heavy here. This one is deliberately simple to stand up, which makes it a strong candidate to run right after quoting proves out: low build, fast signal, and it directly feeds your $10M goal without adding office headcount.

05Implementation Roadmap

Now–90 days: **Stand up the quote assistant (Opportunity 1) as the first proven win.** Start here because it's fast to show value and hits a revenue lever your accounts already told you matters. Connect it to P21 pricing/availability, load your category margin targets and contract prices, and put the one-screen approval in front of Deb, Tony, and Rhonda.
**Turn on quote-outcome logging from day one.** Log every quote with value, account, and won/lost so that within the quarter you have the win rate you currently don't track — that baseline sizes every later decision.
**Make Rhonda the named owner of the quoting pilot.** She's your skeptic-in-chief; if the approval screen survives her, it works. Give her sign-off authority on the pricing rules before anything goes live.
3–9 months: **Build and pilot the order-entry capture (Opportunity 2) on your 2–3 highest-volume PO formats.** Run it in parallel with manual re-keying until extraction accuracy and the part-number match rate hold, then widen. This is funded and de-risked by the quoting win proving the approach.
**Grow the customer-part-number cross-reference table** as a shared asset — every confirmed match makes both quoting and order entry faster, and it's exactly the institutional knowledge you want out of people's heads.
**Launch the dormant-account reactivation queue (Opportunity 3)** as a standing daily rhythm, drafted from P21 history, once inside sales has recovered time from order-entry automation to work the list.
9–18 months: **Move clean, known orders to confidence-gated one-click posting** once the match table has matured, concentrating Deb and Tony's attention on judgment orders only — this is the step that lets order volume grow toward $10M without adding office headcount.
**Automate the monthly margin report** off the same P21 data connection built for quoting, reclaiming your full owner-day each month and beginning to get the reporting knowledge out of your head.
**Begin a collections-assist workflow** that surfaces the $380K aging past 60, ranks it, and drafts approved reminder letters for Rhonda — a natural next build once the earlier wins have proven the parallel-to-P21 model.

06Risk & Mitigation

Single-owner and single-champion dependency

Almost everything runs through you and Rhonda, and Rhonda is already at capacity. If reactivation or quoting oversight lands entirely on her, it becomes 'the thing that can wait' again — exactly how the 2019 blitz died.

Mitigation: Scope each rollout so the recurring work is a short daily rhythm, not a project: a ten-account reactivation list, a one-screen quote approval. Assign the daily working of the reactivation queue to Deb or Tony (or a part-timer) with Rhonda owning only sign-off, so her role is governance, not labor. Document each workflow as it's built so it survives a transition — which directly serves your eight-year goal.

A wrong number or wrong shipment escaping to a customer

Your nightmare is wrong shipments, and an extraction or quoting tool that auto-sends could ship a mis-parsed line or quote a stale price at scale — faster mistakes are worse mistakes.

Mitigation: The human-signature rule is built into every design here: side-by-side review before posting orders, one-screen approval before quotes send, drafted (not auto-sent) reactivation emails. Low-confidence lines are flagged for exactly this reason. Confidence-gated auto-posting is introduced only after the match table has proven itself in parallel, never on day one.

Tool outlives its usefulness if it can't be run without the builder

You were explicit: if you build it, my people run it after you're gone, or it doesn't come in the door. A clever system no one on a sixteen-person, non-technical team can operate is a stranded asset.

Mitigation: Every workflow ships with plain-language runbooks and hands-on training for Rhonda and inside sales, and each is designed around the tools your team already touches — the counter guy already runs ChatGPT comfortably, so the bar is a review screen, not a programming task. We build in parallel, hand over running, and structure the engagement so there's no black box only we understand.

07Next Steps

Pull your real quoting and order numbers for a week. Have Deb and Tony tally, for one week, how many quote requests came in, how long each took, and how many orders they re-keyed by hand — plus every wrong shipment and its cost. You gave me strong estimates; a single week of actuals turns the cost figures in this report into numbers you can bank decisions on.
Name Rhonda as the assessment point person and block her an hour. Put Rhonda formally in charge of evaluating the quoting pilot and give her one protected hour to walk through the current quote and PO-entry steps with us. She's the right owner precisely because she's skeptical; her sign-off is your quality gate.
Export the dormant-account list from P21. Run the list of accounts with no order in the last 90 days, ranked by prior spend, and pull whatever notes you still have from the 2019 blitz on what worked. That single export is the seed for the reactivation queue and costs you nothing but a few minutes to generate.

About This Report

This report was produced from a guided diagnostic interview with you about how Mound City Industrial Supply actually operates — your order flow, quoting, reporting, receivables, team, goals, and constraints. Every finding traces back to something you told us: your 80 orders a day, the 25–30 weekly quote requests, the $380,000 aging past 60, the 900 accounts and 2019 blitz, and your rules about keeping P21, keeping a human signature on customer-facing work, and doing one proven win at a time. The cost figures are conservative estimates built from the baselines you gave, with assumptions labeled so you can swap in your own numbers. Treat this as a starting framework for prioritizing where to act, not a final implementation plan — the next step is validating the baselines above and scoping the first build in detail.