What you told us, what we found, where this goes, and the questions worth bringing to the call.
The week of September 14, we are meeting to discuss three things you raised. Here they are, in your words, before any of ours appears. No solutions on this section.
A tool at the bench that a technician attaches to the device under test, which then simulates what that device would do in a running machine, in order to determine whether it is faulty and what specifically is faulty.
Ideally it recognizes the device and the equipment it belongs to, tractor or combine and so on, without the technician having to look it up.
Three months of overwhelming volume during the busy season (planting and harvest), then a call center with little to do the rest of the year.
The problem is not that the calls are hard to answer. The problem is that the staffing required to handle peak cannot be justified against what the trough actually needs. You want to absorb the peak without carrying the trough.
The bench tool is the starting point. The extension is to your dealers, their field teams, and ultimately farmers in the field.
Concept: a ruggedized tablet that plugs into the machine on site, at the farm rather than back at the depot, and runs the same diagnostic engine, either preloaded or pulling from an Ag Express server when connectivity allows.
You are working with an outside developer on harness cataloging, framed as an AI project. We will come back to that on the next page. The short version: the catalog is the right data. We are not here to replace the work; we are here to show what it becomes when it sits underneath a full diagnostic infrastructure instead of standing alone.
No solutions on this section. What follows is what we found when we looked.
We spent time with the public record before this call. This is what it told us. Every factual claim below is sourced and linked. Inferences are labeled.
Ag Express Electronics handles repair, calibration, distribution, and the Vanguard product line. Integrated Circuit Works (rebranded December 2023 from Alliance Group Technologies, acquired 2020) handles wire harness manufacturing, box builds, and electronics assembly. ISO 9001:2015 certified by NQA, August 2023. PR.com, Dec 2023
North American repair center for Topcon Positioning systems. agexpress.com North American repair center for McHale electronic equipment, announced November 2024. agexpress.com, Nov 2024 Authorized Topcon distributor across Nebraska, Iowa, Illinois, and Indiana.
He joined as a technician in 2002. Became COO. Named permanent CEO September 2024. Business Record (Des Moines), Sept 2024 He will validate or reject the bench concept on technical merit, from personal experience, in real time. The diagnostic bench idea is almost certainly his own long-standing frustration made into a product request. inferred
Their published commitment is 3 to 5 business days. The same page adds that seasonal volume and testing requirements affect timing. agexpress.com, repair service page That caveat is the bench tool's business case in one sentence. The goal is not to remove the caveat by overstaffing; it is to remove the need for it.
Ag Express is an independent repair provider with 34 years of practice fixing what the dealer channel would not or could not. As of two months ago, the largest manufacturer in their market is under a decade-long federal order to hand independent repairers the same software capabilities as its own dealers. The ceiling on independent diagnosis just moved. inferred
The window where accumulated failure knowledge is a durable moat is open now and is not permanent. Knowledge that stays in technicians' heads does not scale into that window.
New farm machinery sales are down 15-20%, dealer profitability at a five-year low, and dealership consolidation underway. Morning Ag Clips, June 2026, citing Titan Machinery CFO Farmers are keeping older machines running for another season. This is a tailwind for Ag Express specifically; their entire product logic is countercyclical to new equipment sales. inferred
BLS estimated 36,830 farm equipment mechanics and service technicians employed nationwide in 2023, against widely reported shortages. farmdoc daily, University of Illinois, Jan 2026 Ag Express's own site lists technicians with 14, 16, and 20 years of tenure and 250 combined years across the team. agexpress.com, technician profiles Every year of that is undocumented, unstructured, and walks out the door at retirement.
Iowa State University Extension puts a single 8-hour day of equipment downtime at roughly $2,400 during planting for 12-row equipment. ISU Extension, farmdoc coverage The value of compressing the diagnose-and-quote step is not primarily technician hours saved; it is turnaround days removed during the 60 days of the year when downtime costs the most.
| Competitor | What it does | What it does not do |
|---|---|---|
| Jaltest AGV (Cojali) | Multi-brand agricultural diagnostics. Fault codes, live data, guided troubleshooting, calibrations. Sold as a kit with rugged laptop and interface. jaltest.com | Talks to an assembled, powered machine through its diagnostic port. Does not work on a loose component on a bench with no machine around it. That is the gap. |
| OEM scan tools (Deere Service ADVISOR, CNH EST, AGCO EDT) | Single-brand, dealer-priced. Now subject to broader availability under the FTC order for Deere. | Single brand. No cross-brand failure corpus. No bench capability. |
| MST-9000+ class signal simulators | Generate crank, cam, sensor signals to make an automotive ECU believe it is in a car. Roughly $400. technician forums, verified pricing | Automotive only. No ag coverage. Built-in database limited to a few dozen ECUs. Users are on their own for anything not listed. |
| dSPACE SCALEXIO / NI HIL | Genuine hardware-in-the-loop. Real ECUs tested against a real-time simulated plant. Fault insertion. 24/7 automated test execution. dSPACE markets into ag and construction mechatronics. dspace.com | Six figures. Engineering staff required. Built to validate a design, not to triage a returned part. OEM validation lab tools, not repair depot tools. |
| Aquant | Agentic AI for organizations that build, sell, and support complex equipment. Converts manuals, service histories, parts and warranty data into agents serving field techs and contact center agents. Customers include Terex, Powerscreen, Comfort Systems. aquant.io | SaaS. The corpus is processed on their infrastructure. They have already proven the thesis; that is not a problem, it is validation. The difference is sovereignty, scope, and fit (see below). |
| Neuron7 | Service resolution intelligence built on a Service Expertise Graph derived from case history. Guided resolution, triage, enterprise search. Integrates with Salesforce, ServiceNow. neuron7.ai | Targets Fortune 1000 enterprises with 1,000+ technicians. Ag Express has roughly 175 employee-owners total. They are structurally too small to be a priority account, and they will be sold a configuration of somebody else's ontology. |
The gap, stated plainly
Between a $400 automotive toy and a $250,000 OEM validation rig, there is nothing built for a multi-brand ag electronics repair depot. Filling it does not require the best simulation technology. It requires the best failure corpus. Ag Express has been building that corpus for 34 years. dSPACE does not have it. Aquant cannot have it. Only Ag Express can have it.
The three-axis differentiation from Aquant and Neuron7: sovereignty (their corpus on their hardware, not on a vendor's cloud), scope (the whole company, not just the service department), fit (built for their size and their industry, not for Fortune 1000).
agexpress.com runs on WordPress with Google Site Kit. solutions.agexpress.com runs on HubSpot CMS. store.agexpress.com is a separate e-commerce storefront. page source, rendered metadata Three platforms managed separately, none of them connected to the others.
A repair catalog routing fault on store.agexpress.com returns a redirect loop. observed, September 2026 The site also notes that the 2026 spring product catalog is still under construction. agexpress.com, observed September 2026
This is not offered as criticism. It is the most concrete, non-speculative, immediately visible demonstration of what we are proposing. The catalog should be generated from the same knowledge graph that holds the parts, the repair services, the machines they fit, and the failure modes they address. The same graph feeds the store, the campaign pages, and the bench. Fixing the catalog is the easiest first node, and it is already visibly broken.
This is the part we built last. It is built on your own thesis.
"Farm equipment stays mechanically viable for twenty to thirty years while its electronics remain serviceable for only three to seven. That asymmetry drags on resale value."
You have stated the problem. The corpus solves it.
Nobody else can. Here is why:
No bench hardware required. No tablet. No simulation engine. No dealer network. Just the corpus, which Phase 1 and 2 build anyway. The data product is a byproduct of a corpus that already has to exist for the bench to work.
| Buyer | What they buy | Why now |
|---|---|---|
| Auction houses and equipment marketplaces | An electronics risk score on listed machines | Differentiated listings. Buyer confidence. A defensible premium on well-scored machines. |
| Equipment lenders and lessors | Residual value risk on the electronics component | They currently price residuals on mechanical condition and hours. They are blind to the thing that actually degrades first. |
| Insurers | Failure probability by model year and system | Underwriting input that does not exist anywhere. A new underwriting line with no legacy competition. |
| Dealers taking trades | Risk on the unit before they price it | Dealer consolidation and used-inventory pressure in the 2026 equipment trough make this acute right now. |
| Farmers buying used | A pre-purchase electronics report | Direct consumer product. High volume, lower price point. No current alternative. |
New farm machinery sales are down 15-20%. Farmers are keeping older machines running for another season. More used machines are changing hands. More people need to price them. The repair business is already countercyclical to new equipment sales; their entire reason to exist is that old electronics fail.
The residual-risk data product amplifies that positioning in a direction that has no ceiling. The bench tool makes repair more efficient. The data product monetizes the knowledge directly, without requiring a dealer network or a tablet or a simulation engine.
It is plausibly larger than the bench tool. It does not require Ag Express to become a software company or convince anyone of anything. The buyers already exist and are already pricing the question badly.
Phase 4 includes the external distribution of this data product. Whether that distribution lives inside Ag Express, through a dealer agreement, or in a separate legal entity is a structure question, one for your board and your counsel, when the corpus has earned the conversation.
We are not proposing a structure here. We are showing you where the corpus goes if you build it. The structure follows the business, not the other way around.
Phase 4 does not start on a schedule. It starts when a lender or auction house asks what an electronics risk score would cost. That request is the market signal. Its absence is also a signal.
23 questions, ranked by what each answer unblocks. Bring no more than eight to the call. The portal is doing most of the talking.
This question unblocks everything. If the answer is a system name with structured fault codes, we are a year ahead of schedule. If the answer is "in the ticket notes" or "in his head," Phase 1 becomes structured capture first and the timeline moves.
Follow-up: how far back does it go, and can it be exported.
The top five device families by volume are the only families worth building Rung 3 for. Without this number we are guessing at where to point the engineering. And the month-by-month shape determines whether rung 2 trace capture is worth building per device family.
Ask Randolph specifically. He came up through the bench. This is the question where he stops evaluating us and starts collaborating, and it is worth spending real time on. What is already there shapes what we are augmenting versus what we are starting from scratch.
The posture here is genuinely curious, not competitive. That catalog is the right data. A structured catalog of harnesses, connectors, pinouts, and the machines they fit is precisely the substrate the diagnostic engine needs. We want to absorb it, not replace it.
No dealer locator was found on their public properties. Dealer count directly sizes the field-tablet opportunity, the eventual licensing model, and the residual-risk data product's distribution network. Field-of-use restrictions in dealer agreements may also constrain what Ag Express can build or sell around their relationship.
Unblocks: VOICE-07, ROI on the intake node, and whether season one needs a commodity vendor instead of a built one.
Reveals whether the bottleneck is call capacity or expertise depth. Different problems, different solutions. Also unblocks VOICE-10 escalation design.
Unblocks installation logistics, remote access posture, and who our day-to-day counterpart is. Whether on-prem is a comfortable idea or an alarming one matters to the sequencing.
We know they run WordPress, HubSpot, and the storefront. What runs behind those is the ingestion surface for CC-07. The answer changes the integration scope significantly.
A CEO can sponsor this. Only an engineer can validate the bench claims. An engineer who feels bypassed will kill the program quietly. Ask early.
Two months old, ten-year term, under FTC and five-state supervision. Independent repair providers get the same software capabilities as authorized dealers. He may not have connected it to his own strategic position yet.
Offered as help, not as a finding. It is the most concrete, non-speculative demonstration that we did the work, and it has nothing to do with AI.
You are entering harvest. The phones are about to be at their loudest and the benches at their fullest. This is simultaneously the best time to make the argument and the worst time to ask for their attention. The first sixty days are ours.
The model has your variables blank. We will not put a total in the empty cell. Filling them is exactly what the discovery engagement is for, and a vendor who shows you confident numbers built on assumptions you can immediately falsify is not a vendor to trust with your building.
Verdict sign-off, calibrated confidence language, counsel-drafted disclosure, insurer review before Rung 1 ships. This conversation happens before the first customer sees a result, not after. We bring this up.
Hardware, code, corpus, models, runbooks. Not licensed back to you. Not on our platform. Purchased in your name, running in your building, transferable without our involvement. That is not a concession. It is the whole differentiator.
Four to six weeks. Fixed fee, credited in full against the first retainer month if you proceed within thirty days. Everything diagnostic, nothing constructive. This is what the next call is asking for.