Cadmon Systems builds AI agents that read emails, contracts, PDFs and WhatsApp messages, apply your business rules and write structured data into the ERP. A person approves every sensitive write, and Make.com carries out the rest inside your systems.
20+
Pages in a single contract the AI analyzes
25 to 30
Contract templates in several languages
5
Priority environments receiving the data
Pinecone
Knowledge base for answers from your own documents
Contracts, requests for quotation, emails and inquiries each arrive in a different format. Someone reads them, finds the data and types it into the ERP. An AI agent does the reading and the extraction, and the data reaches the system as a draft for approval.
✕ Long contracts and documents read and typed in by hand
✕ Repeat questions on WhatsApp and email wait for an agent
✕ Information in company documents is hard to find when you need it
✓ AI extracts the fields, and Priority receives a draft for approval
✓ An AI agent recognizes the type of inquiry and routes it
✓ Ask a question and get an answer from the documents themselves
Agents that work inside the processes you already run, not on the side.
DocuPipe parses contracts and invoices into fields, including multilingual documents of 20+ pages.
A request for quotation from email, PDF or Excel becomes a quote in the ERP, and sensitive lines go to a person.
An agent that recognizes the type of inquiry and routes it to the right sub-agent, for example by branch.
RAG with Pinecone: ask a question and get an answer from your own company documents.
The AI applies your business rules, escalates exceptions to a person, and a person approves before the ERP write.
An assessment for leadership: where AI will save work in your company, and what to prepare before you start.
We work in phases, and you see a result at every step before we move on.
01
A written document covering which documents or inquiries come in, which fields we extract, the rules, and when a person approves.
02
One complete flow on real documents. The data arrives as a draft, and a person approves before the ERP write.
03
We sit down with the person who owns the process and check what the AI gets right and what needs improving.
04
We add templates, document types, channels or branches, and you get clear documentation of everything we built.
The challenge: A global automotive supplier received customer price contracts in 25 to 30 templates and several languages, and entering one contract by hand took up to several hours.
The solution: Contracts arrive at a dedicated mailbox. DocuPipe parses them against a unified schema, and an LLM layer analyzes free text in documents of 20+ pages. Make.com matches customer and part codes, and Priority receives a draft for a person to approve.
Contracts reach five Priority environments as drafts: Canada, the USA, Germany, Tianjin and Spain.
More projects where an AI agent does the reading and a person approves what matters.
A WhatsApp AI agent for a parking technology company, connected to monday.com.
Data extraction and RAG over lease contracts for a US property management company.
Requests for quotation are read automatically and arrive structured, ready for pricing.
The data the AI extracts is written to Priority, including custom forms and subforms.
Five connected services: Make.com, business process automation, AI agents, Priority ERP and monday.com.
Can AI extract data from contracts into an ERP?
Yes. Document AI such as DocuPipe parses contracts into structured fields, and Make.com writes them to the ERP as drafts for human approval. Cadmon Systems runs this for a global automotive supplier across 25 to 30 contract templates and five international Priority environments.
What is an AI RFQ agent?
An AI RFQ agent reads incoming requests for quotation from email, PDF or Excel, extracts the line items, applies pricing rules, and creates a quote in the ERP. Cadmon Systems builds RFQ agents on Make.com, DocuPipe and the Priority ERP API, with high-value or sensitive lines escalated to a person before the quote goes out.
How do you keep AI from putting errors into the ERP?
The AI follows your business rules, sends exceptions to a person, and a person approves before any data is written to the ERP. The data arrives as a draft and becomes a record in the system only after approval.
What is RAG, and why does it suit company documents?
RAG (retrieval-augmented generation) answers questions from your own company documents instead of the model's general knowledge. Cadmon Systems builds RAG with Pinecone as the vector store, for example over the lease documents of a US property management company.
What is iPaaS as an AI execution layer?
iPaaS as an AI execution layer means AI models decide, and an integration platform like Make.com executes those decisions safely inside ERP, CRM and other systems. Asaf Cadmon presented this approach at the 2026 CIO Club conference.
On a discovery call we go through the documents and inquiries that reach you, and decide together what to hand to AI first.
