AI Agents and Software for Contract Manufacturing
A request arrives by email: three drawings as PDFs, 250 parts in 1.4305 stainless steel, delivery in four weeks. Someone has to read the drawings, check material, tolerances and surface finish, choose the right machine, look at capacity, calculate and reply. Often that is the same skilled person who also writes the programs or sets up the machines. If the request waits two days, the job may go to the shop that answered faster.
What an AI agent can prepare for a request
An AI agent takes over the reading and groundwork before a person calculates. It works with the knowledge your company already has and ends with a draft.
- Capture the request: parts, quantities, material, delivery date and the details from the title block and drawing notes, such as tolerances, surface finish or heat treatment.
- Name the gaps: find missing or contradictory details and word the questions for the customer.
- Compare with past jobs: find similar parts, how they were calculated and the machines they were made on.
- Draft the quote: transfer line items, assumptions and open points into your quotation template.
For that comparison the agent needs access to your job history and your machine park, whether they live in the ERP, a calculation spreadsheet or job folders. That connection is often the larger part of the work. Where your existing programs do not provide it, we build the missing software: the data access, the approval step and a log that shows where each detail in the draft came from.
What stays with you
A person decides the price, the delivery date and the commitment to the customer. Your team knows whether a part pays off on your machines, whether the machine is free in four weeks and which customer has priority right now. The agent proposes and explains; your estimator checks, edits and approves.
No quote leaves the company without that approval.
Where our knowledge of contract manufacturing comes from
No contract manufacturer is among our clients yet. What we know about the industry comes from two sources, and both are market knowledge, not project experience inside a contract manufacturing shop.
For MAFU-SHERPA, a manufacturer of AI-controlled CNC automation for small and medium batch sizes, CERTAINCE built an AI sales assistant that runs in production there. It reads the websites of manufacturers with CNC turning and milling machines and evaluates machine park pages, job ads for CNC specialists and signs of small batches or new machining centers. To weigh those signals, we had to understand how such companies present themselves and how to tell that they are investing.
For our own sales, AI agents research companies every week from a pool of roughly 1,900 exhibitors at three trade fairs for metalworking and the supplier industry: AMB, Z Zuliefermesse and Intec. In that work we read how suppliers and contract manufacturers describe their services, their machine park and the way they take requests.
We therefore cannot show you results from a contract manufacturing shop; whether the workflow holds up in yours, we test on your own requests.
The first step
It usually starts with an opportunity assessment. In one day we go through up to two workflows with you, for example request and quote, each on a few real examples from your company. The report afterwards says which tasks an agent can take over, what your data allows today and whether a paid next step is worth it. The answer may also be that none is worth it yet.
If it is enough for your team to hand requests to Claude or ChatGPT themselves, the setup with an AI enablement workshop is the next step. If the agent needs access to ERP and job data without a ready connector, or your calculation rules need owned software, that comes with automation; the code is then yours, tested and maintainable.
If you want to go through this on a real request, let us discuss a concrete workflow.