Bill of materials and sourcing

Real part numbers, alternates and honest costs

Your AI turns a design into a BOM a manufacturer would accept, the way a sourcing lead would: specific part numbers in production, an alternate for every risky line, costs at 1, 10 and 100 units, and the supply risks named.

By Vander Labs. Works with Claude, ChatGPT, Cursor and Claude Code.

What your AI learns

• Chooses specific manufacturer part numbers that are still in production
• Adds a drop-in alternate for every non-commodity line
• Exports a dated, costed CSV and a list of supply risks

How it works

1. Start from the design: Walks the block diagram so every function, connector housing and screw becomes a line.
2. Choose parts: Manufacturer part numbers, never generic descriptions; unknown numbers are marked verify, not invented.
3. Selection rules: Active production, two distributors, pre-certified radio modules, capacitor voltage margin, hand- or machine-friendly packages.
4. Alternates: A true drop-in for every non-passive line, or a note on what changes.
5. Costs: Per-unit totals at 1, 10 and 100 with attrition, and PCB, assembly and enclosure as separate lines.
6. Sourcing: Authorized distributors for chips and safety parts; marketplaces for prototype modules, with counterfeit risk flagged.
7. Price check: Every price dated and sourced or labeled an estimate; nothing from memory presented as a quote.
8. Compatibility: Logic levels, regulator headroom, module voltage ranges, connector mating and cable lengths checked before finishing.

Limits and safety

• Never invents a part number; unverified lines are tagged with the search needed to confirm them.
• Every price is marked as an estimate with a date unless it was looked up and cited.

Topics: bom, sourcing, components, digikey, lcsc