ABOUT EDMA

Built in production at EDMA Group. Two years live. Then opened up.

EDMA Group is a physical-trade business — 9 manufacturing countries, multiple distribution markets, roughly 20M boxes a year. TradeOS runs the operation. We didn't ship it as a product until two years of running it ourselves had proved out what mattered and what didn't. This page is that story, and the architecture it produced.

Read the storySee the architecture

GMV

$80Mannual · physical trade · EDMA Group

Operating record

2 yrcontinuous dogfood · same code in prod

Manufacturing

9countries · real factories · real lots

Volume

~20Mboxes per year · multiple markets

Tenants today

1just us · until you

HOW WE GOT HERE

Most trade software is built by people who've never run a trade. We're the inverse.

EDMA Group started as a physical-trade operation. Over time we expanded to 9 manufacturing countries and multiple distribution markets — and our software stack grew to 14 tools held together with email and spreadsheets. NetSuite for the ledger. SAP for the supplier side. Salesforce for the buyer side. DocuSign for the contracts. Slack for operations. Email for everything in between.

The compound cost of moving information between those tools was eating roughly a third of our trade-ops headcount. People weren't doing trade — they were transcribing data between systems that didn't know about each other.

So we built TradeOS. First for ourselves, then for the operators we knew were paying the same tax. Every section in the product was built because we needed it on a deal that day, not because it looked good on a roadmap.

THE TEST WE HELD OURSELVES TO

The version you see at edma.trade is the public release of the same platform EDMA Group runs on internally. No fork. No special "demo" build. No separate enterprise version. When something ships to us on a Tuesday, it ships to operators on Wednesday. When something breaks on a real deal, it gets fixed on a real deal.

OPERATOR-FIRST DECISIONS · PROOF POINTS

Things only an operator would put in the product.

Six design decisions in TradeOS that we wouldn't have made — and probably wouldn't have noticed — if we weren't running real deals on the platform every day.

01

The Risk panel has 6 signals, not 12.

We tested more in the early designs. The financier reviewer never read past six. Six is what fits on a screen without scrolling and what an underwriter holds in their head at once. We removed the rest. Six is now the locked spec.

→ Buyer credit · Assignment status · Insurance binder · Document completeness · Production & shipment · Client claims

02

Disclosure is two-step, not one.

A financier that signals interest on a listing doesn't see the operator's name until the operator approves the financier's firm. Operator and financier reveal in the same moment, by mutual consent. The default in most marketplace products is one-step (you bid, they see you). We took the operational hit (slower matching) for the protocol gain (identity protection until commitment).

→ listing.disclosure_state = 'pending' | 'disclosed'

03

The marketplace fee is on the financier, not the operator.

The platform fee is charged to the financier. Operators pay a platform subscription to access TradeOS — they don't pay a separate listing or matching fee on top of that. Charging the operator a deal fee would re-introduce the very friction we built the marketplace to remove.

→ Financier-side: platform fee per repayment event · Operator-side: subscription only

04

The Settlement Waterfall computes itself.

No spreadsheet. Buyer pays per the Notice of Assignment, the financier deducts capital + fee + platform fee, the operator residual auto-remits T+2. We wrote the waterfall logic the first time we factored a real receivable on our own platform and realized we were about to do the math in Excel like everyone else.

→ 7 deterministic rows · contract-derived · zero reconciliation

05

Documents have visibility tags, not "shared / not-shared."

A bill of lading is shared with the client. A factory invoice is not. A supply agreement is shared with the financier but not the client. Five visibility states across six counterparty types — 30 distinct exposure permutations across the document chain. Off-the-shelf collaboration tools have two states. We built our own because we had to.

→ Operator · Supplier · Client · Logistics · Financier · Platform

06

The client portal renames "Invoices" to "Payments" on mobile.

Clients think "I'm paying things," not "I'm processing invoices." We renamed the mobile-nav route after our own clients couldn't find the bills. Same data, different label. Vocabulary is product.

→ Same route · two labels · UX wins

THE TEAM

Small team. Long memory. Bias toward shipping.

The platform is built by a team that has run physical trade. Nick, the founder, designed the system and operates it every day on real EDMA Group deals across nine manufacturing countries. The engineering team is small and works in tight loops with the trade-ops team — the people whose Mondays get easier when the right thing ships.

There are no consultants, no agency engineers, no white-label code. Every line of TradeOS was written by someone who has either traded the physical product or watched the people sitting next to them trade it.

NEGATIVE SPACE · USEFUL TO STATE

What TradeOS is not.

Most product pages tell you what something is. The faster way to convey what we're building is what we've decided we're not.

× NOT

A marketplace play.

The marketplace is one tab of TradeOS. The platform's value is the operating record, not the matching. If we never matched a single deal between an operator and a financier, the orders / production / shipments / documents / tasks / finance / dashboard sections would still be a complete operations stack.

× NOT

A vertical SaaS pretending to be horizontal.

We started in physical trade. Every word of the data model assumes a real cargo moving between real parties across a real border. We are not interested in being the next ServiceTitan, Procore, or Toast. The horizontal expansion — if it ever happens — is into the network, not into other industries.

× NOT

Fintech.

The marketplace is operational infrastructure with a financing surface attached. We charge a platform fee, not a spread. In v1 we don't hold money — the financier pays the supplier direct; the client pays the financier direct. We are not a bank, not a market maker, and not a custody platform.

× NOT

VC-led.

EDMA Group bootstrapped through trade margin, not term sheets. The platform was funded by the physical-trade business that needs it. Outside capital, when it comes in, is sized to extend the operating record and the build cadence — not to chase logo deals or hit a growth-at-all-costs curve.

× NOT

"AI-first."

The product has AI in it — Atlas Document Intelligence parses the document chain, the Accounting AI reconciles payments against invoices, the Bot Studio lets operators build agents in plain English. All of that exists because we needed it on a deal that day. We are an operating record that uses AI where it helps. We are not a chatbot looking for a workflow.

WHERE THIS IS TODAY

Two surfaces in production. No vaporware on this page.

We say what we ship and we say what we don't ship yet. The list below is the part of the platform that's either running EDMA Group today or in active build for public v1.

  1. 01

    TradeOS

    The operating record. 19 sections, 4 portals. Runs the entire physical-trade lifecycle — orders, production, shipments, documents, tasks, finance, dashboard, communications — for the operator and their counterparties.

    Shippedrunning EDMA Group · public v1 June 2026
  2. 02

    Trade Marketplace

    Pre-vetted financiers bid on listings the operator posts. Platform fee paid by the financier per repayment event. Two-step disclosure protects identity until both sides agree.

    Operator side shippedfinancier portal in progress

Beyond these two surfaces, we've scoped longer-horizon architecture work tied to settlement rails and capital-markets adjacencies — but those don't get marketed until they ship. The discipline is that the only things on this page are the things you can actually run a deal on today, or will run a deal on in v1.

About — Built by trade operators, for trade operators | TradeOS