Diogenes deconstructs your parts the way a cost engineer would: ground up, physics based, every data point traceable. Then it hands you a number that holds up in a negotiation.
The old way needs a CAD seat, a trained cost engineer, and weeks per part. The study is stale the day it's done. Diogenes runs on the ERP spend file you already have.
The ERP extract you already have: parts, prices, volumes, suppliers. No CAD required to start, no IT project, nothing to integrate.
Each part is routed to credible processes and priced from the physics up, with cycle times from geometry and rates from cited public data.
Walk in knowing the supplier's cost better than they present it: line-by-line breakdowns, live levers for process, region, and volume.
Not one part at a time. Diogenes prices every line in the file, compares each against what you actually paid, and sorts your spend base by recoverable dollars, so you know where to spend the next negotiation before you open a single drawing. Every line opens into the teardown below it.
Below: five years of purchase history from a mid-market fluid-power maker, read in one pass. $890K a year, on a $16.5M spend base, sitting in sixteen parts. A cost-reduction program does not need to touch the other fourteen. And a purchasing dollar is not a revenue dollar: it lands on EBITDA in full, so at a 10× multiple this one file is worth roughly $9M of enterprise value.
Diogenes reads the solid, measures the removal volume and features, and costs the part under every manufacturing route it could plausibly take. Feasible routes get ranked and priced. Infeasible ones get ruled out, with the reason shown. Here is a benchmark part it read from a raw STEP file — hover any marker on the solid to see the feature the analyzer found there, its dimensions, the machining operation it drives, and how confident the engine is that it read it correctly:
$6.21 per unit over should-cost, at 15,000 units a year. That gap, line by line and every element cited, is your negotiation.
Should-cost vs price paid, line by line. When the supplier offers new information in the room, their region, their crewing, their machine utilization, their scrap, you requote the number in seconds. Machine burden is rebuilt for each region from floorspace rent, electricity, depreciation, capital, and OEE, so the machining line moves when the plant moves. Two reference lines anchor the conversation: the nominal should-cost, and an achievable floor recomputed at best-run assumptions. Every figure below was computed by the Diogenes engine on the valve block above, from the live rate database:
$6.21 per unit over should-cost, at 15,000 units a year. Every element cited.
Two suppliers, two regions, each priced against its own regional should-cost. The dashed floors below were computed by the engine on period-correct rates, quarter by quarter back to 2021, including the 2025 duty change you can see hit the German line. The incumbent has pulled above its floor, and the gap keeps widening. The gap is the negotiation.
Ex-works price is only the start of the argument. Diogenes builds the full landed comparison for the same valve block, with freight lanes and duty modeled from the USITC tariff schedule, and flags overpaid lanes for clawback. China is the cheapest plant and the duty takes most of that back:
A should-cost only works if every input survives scrutiny. So every input is built to.
Machine rates, labor, metal indices: each traces to a named public source with an as-of date, shown beside the number it feeds. Nothing is a house constant.
Cost models are benchmarked against independent industry references, process by process, and refined where they diverge. Where a model is still weak, it says so.
Add materials, whole process models, or parametric archetypes for the spend with no CAD. Author them by hand or have the LLM assistant draft one, then calibrate it before it goes live.
22 purpose-built process models, each costed from the physics of that process, plus the secondary operations that quietly carry the cost: heat treatment and finish machining are modeled, not marked up. And it is extensible: add your own processes and materials, and they route alongside the built-ins.
Costing tools exist on a spectrum, and each is built for a different job. Diogenes is built for operators and investors who start from a spend file and need leverage fast.
CAD goes in, cost comes out, wired into PLM across a global engineering org. Powerful when every part has clean CAD and a cost-engineering team runs the tooling.
An engineer describes the part, explores tolerances and simplification, and reworks the design. Powerful when you own the design and have the engineering hours.
Starts from the ERP spend file you already have, ranks the entire spend base, and arms the supplier conversation with cited, graded should-costs.
Teardown-based should-cost works. The problem has always been that running it takes a lab, a trained cost engineer, and weeks per part, which puts it out of reach of the companies that need the leverage most. Diogenes exists to close that gap.
Cost falls out of geometry, material, and the machine that has to cut it. That is why Diogenes models the physics instead of reading a number off a lookup table.
An estimate is only worth what it survives. A number a supplier can wave away across the table is not leverage, however elegant the model behind it — so every element here is built to be defended line by line.
Led the global Design to Value and Teardown lab network: taking products apart, across industries and regions, to find where the money actually sits.
Diogenes is that method made repeatable — the same teardown discipline, run against the spend file you already have instead of a lab bench, with every number cited and graded.
Direct material is your biggest controllable cost, and should-cost is the most credible lever in a supplier negotiation. Three ways in:
A fixed-scope read on one spend file: where you're overpaying, by how much, and which dollars are realistically recoverable, split between re-source leverage and negotiate-in-place.
We run the negotiation cycle with your team: teardowns, supplier packs, branded leave-behinds, with fees at risk and paid on realized savings.
Deploy Diogenes inside your firewall and run it continuously with your own people: every quarter, every new quote, every portfolio company.
The diagnostic starts with the ERP extract you already have. If the money isn't there, you'll know that too. Every number cited and graded.
Request a diagnosticPrefer to start smaller? Send one part and get a should-cost teardown back.