The Indax manifesto

Preface

Every machine, every plant and every product in the United States is a chain of orders between companies. A packaging line is an OEM's purchase orders to forty suppliers, each supplier's purchase orders to its own, and a promise date at every link. US manufacturing moves $7.2 trillion a year and today it is all coordinated through email.

• Every RFQ, quote, purchase order, promise date, product spec, bill of materials, certificate and invoice crosses every company line as a PDF or Excel file attached to an email.
• A person on the other side reads it and types it into another system.
• A reply takes on average five days.
• Sales and purchasing teams send upwards of 200 emails per day to coordinate.
• Delays across a supply chain take weeks to reach the end customer, creating unnecessary risk to production all across the world.

Industry has accepted these inefficiencies as the cost of doing business, but this is rapidly becoming a massive bottleneck, especially in the age of AI.

But why?

If we think about this problem from first principles, we hold the following facts to be fundamentally true:

1. Companies across a supply chain share the same primitives when transacting with each other (quote, spec, BOM, order, certificate, delivery, invoice).
2. The information about these primitives exists inside each company’s email, ERP and CRM, but is fragmented and has different schemas across companies.
3. Companies constantly need to share information on these primitives.

Systems of exchange

Indax is a new category of software that allows companies to share context between each other, to accelerate the transfer of information necessary for commerce. Here are the foundational building blocks:

• One shared ontology: to facilitate the exchange of information between these companies we must standardize the data schema around the primitives that they use daily.
• Automatic data collection and resolution: a system of exchange reads company data like email, CRM and ERP, and abstracts the relevant information into the standardized ontology.
• Permissioned between companies: once schemas are shared, information can be synced across the relevant primitives.

A system of exchange is a next-generation EDI. It allows companies not only to send purchase orders or invoices, but also to permission real-time information across all primitives that affect the commercial process.

Why does this matter?

1. The true bottleneck of the physical economy is the time information takes to cross multiple company lines.
2. Agents will inevitably run purchasing and sales in industrial companies across supply chains. Without shared rails, each of those agents stops at the edge of its own company and waits for answers just like a human would, on email.
3. The system of exchange will become one of the most critical technologies for the functioning of the world, just like the clearinghouse is in payments and in healthcare. If the system of exchange fails, agents and humans across purchasing, sales and shipping will have to fall back to manual work (email), collapsing entire supply chains.

Why now?

1. Agents can read the documents. Language models now extract structured records from emails, PDFs and spreadsheets and resolve the same entity across systems. In Indax's September 2026 benchmark, the Indax ontology answered 93% of questions where documents contradict each other, against 40% when answering from the raw files.
2. The bottleneck has moved to production. AI hardware design is advancing quickly, and the constraint is shifting to sourcing, manufacturing and delivery.
3. Agents now have a standard interface. MCP gives every agent one way to reach a data source, so a shared record can serve every agent on both sides of an order.

Why us?

The system of exchange for industrials must be built by people who deeply understand entity resolution and agents, who have worked inside industrial companies and who have sold to the largest enterprises. We grew up inside manufacturing businesses, unified massive datasets, built agents for 4+ years, trained tabular transformers, sold more than $300 million of industrial equipment to governments and scaled production lines to 500,000 units a year.