🤖 AI & Copilots
AI copilots that understand electronics manufacturing
VivyaWorks AI is not a chatbot on the side. It is a set of domain-trained copilots embedded in the workflows where decisions happen. Every copilot operates under tenant-defined rules, logs every action, and never changes data without human approval. Think of it as a very capable analyst sitting next to every planner, buyer, and quality engineer.
RFQ Assistant
Extract, validate, and compare RFQ packages in minutes, not days.
Extract and validate RFQ packages
Parses customer BOMs, drawings, and specs; flags missing data, ambiguous MPNs, and scope gaps; produces a clean RFQ summary for estimation.
Multi-BOM comparison
Highlights delta between BOM revisions so engineers see what changed without line-by-line checking.
Risk pre-screen
Marks known lifecycle, lead-time, and compliance risks at intake so sales can set expectations early.
Compliance Copilot
Answer 'Can we ship this product to this market?' with evidence.
Part-level compliance check
Queries material declarations, substance databases, and tenant rules to produce a Red / Amber / Green status per part.
Product compliance summary
Aggregates part-level results into a customer-safe document showing overall status, exceptions, and evidence gaps.
Regulatory Q&A
Answers natural-language questions like "Is this product REACH-compliant for the EU market?" with traceable references.
DFM Copilot
Plain-language manufacturability feedback before production starts.
DFM risk narrative
Translates DFM/DFT scores and metrics into a plain-language summary highlighting key design risks.
Remediation suggestions
Links each issue to a recommended design or process change, including estimated effort and impact.
Quote risk feed
Surfaces DFM risks inside the quote workbench so cost engineers can price for complexity.
Planner Copilot
Help planners simulate, decide, and promise with confidence.
Promise-date assistant
Calculates realistic delivery dates based on current inventory, open POs, capacity, and lead times.
What-if simulation
Lets planners test scenarios (e.g., expedite a PO, swap a supplier, defer a WO) and see downstream impact.
Escalation alerts
Flags material or schedule conflicts before they hit the shop floor, with recommended actions.
Agent Catalog
Every copilot is backed by one or more agents. Here is the current catalog.
| Agent | Trigger | Task |
|---|---|---|
| rfq-intake-agent | New RFQ arrives | Parse, extract, validate BOM and spec |
| bom-delta-agent | BOM revision uploaded | Compare revisions, flag deltas |
| compliance-check-agent | On demand / gate trigger | Run part-level compliance against regulations |
| dfm-score-agent | Design data imported | Compute DFM/DFT metrics, generate narrative |
| promise-date-agent | Sales asks for date | Calculate buildable date from live data |
| what-if-agent | Planner opens simulator | Model schedule & material scenarios |
| ncr-triage-agent | NCR created | Classify defect, suggest disposition |
| quote-risk-agent | Quote finalized | Summarize risk, flag outliers |
Architecture & guardrails
Multi-tenant isolation
Every model call carries tenant context. Data never leaks between tenants.
Approval before action
Copilots suggest; humans confirm. No agent changes a quote, PO, or schedule without a click.
Full audit trail
Every agent run is logged with input, output, model version, and user who triggered it.
Pluggable models
Swap between cloud LLMs and on-prem models per tenant policy—no code changes required.
See the copilots on your data
Bring a BOM or RFQ, and we'll show you how the RFQ Assistant, Compliance Copilot, and Planner Copilot handle it end to end.
Try copilots on my data