GenAI-assisted RFQ handling put a customer-ready machine concept on the table in days and cut the road to first prototype from 22 weeks to 14.
Figures indicative; validated metrics available on request.
The robotics team builds special-purpose assembly stations to customer RFQs. Most of the delay sat upstream of the workshop, in clarifying, visualising and costing the machine.
Part drawings, takt targets and safety asks arrived across emails and PDFs. Clarifying them took two weeks of back-and-forth.
Customers couldn't see the station until detailed design. Layout and ergonomics feedback came late, forcing redesign.
Press, controls, safety and traceability content was estimated per quote. Quotes slipped, and margins carried hidden risk.
Poka-yoke, MES links and safety interlocks were sometimes confirmed only at trials, adding rework to the first prototype.
GenAI went where the team lost time: reading RFQs, drafting specs and showing the customer a machine early.
Select a step to see what happened.
We walked recent station projects from RFQ email to first trial.
Deliverable: RFQ-to-prototype value-stream map with stage timings.
We made each new station a combination of known building blocks.
Deliverable: Intake template, module library, concept-sheet format.
A repeatable GenAI workflow inside the AC manufacturer's Claude Enterprise workspace.
Deliverable: Prompt pack, concept-sheet workflow, costing model.
Started with one station family, where modules were most standard.
Deliverable: Pilot timelines and refined prompt pack.
The library grows with every project.
Deliverable: Library governance and rollout plan.
A typical servo pressing station, from RFQ to first prototype. Toggle to compare timelines; select a stage to see what changed.
Open item flagged for the customer: press capacity, to be finalised from load data.
Quotes became more complete and arrived on time. More RFQs moved through to orders.
Engineering review score out of 10, sampled quotes
Before (left) vs after (right)
Figures indicative; validated metrics available on request.
A one-page concept sheet surfaced layout, safety and traceability questions weeks before CAD.
A standard intake and a module library gave the AI something precise to work from. Engineers stayed the final check.
Once the concept was agreed, presses and safety parts could be ordered in parallel with design.