01 Product
Stampwise, a quotation agent
for pressure vessel fabricators.
Our first product, Stampwise, is an AI quotation and engineering agent for ASME pressure vessel fabricators. It reads an RFQ package, extracts every requirement line by line, flags gaps and drafts a reviewable quotation workbook. In daily use on live bids at a U stamp fabricator in Korea since early 2026.
01 / INTAKE
RFQ package reading
Specifications, data sheets, drawings and vendor forms go in. The agent extracts the requirements line by line and keeps a source reference for each one.
Useful for Turning a long RFQ package into a requirement list a reviewer can check in an hour.
Long context · Document extraction
02 / GAPS
Gap and clarification lists
Missing data, conflicting revisions and unusual conditions are flagged before pricing starts, with a draft clarification list for the customer.
Useful for Catching a missing design pressure or an impact test requirement before it costs money.
Agent workflows · Validation rules
03 / WORKBOOK
Reviewable quotation workbook
The agent drafts the quotation workbook: materials, fabrication, NDE, testing and quantities, each line traceable. The estimator reviews and signs off.
Useful for A consistent first draft on every bid instead of a blank spreadsheet.
Python · Excel · Deterministic pricing gates
04 / ONTOLOGY
Every number traced to its source
A Neo4j ontology links requirements, components, calculations and prices, so any figure in the quotation can be traced back to the clause or drawing it came from.
Useful for Answering where a number came from during customer review.
Neo4j · Ontology · Knowledge graph
05 / ENGINEERING
Deterministic ASME checks
Thickness and allowable stress checks run in deterministic code with source approved tables, never inside the language model. The agent prepares the inputs and explains the result.
Useful for Keeping ASME VIII Division 1 and 2 checks repeatable and inspectable.
Python · MCP tools · Source approved tables
06 / HUMAN SIGN OFF
Built for high stakes outputs
Pricing and engineering decisions stay with people. The agent proposes, flags and documents; a named reviewer approves every quotation before it leaves.
Useful for Using AI on bids without giving up control of the final number.
Review gates · Audit trail · Agent SDK