Quoting
Intelligence
that compounds.
Nyotta AI reads the full RFQ package, 2D and 3D mechanical drawings, CAD, and spec packs, extracting dimensions, GD&T, and tolerances. It then drafts the quote: a scoped BOM, should-cost estimate, and proposal your engineers approve before it goes out. A fleet of specialized agents that turns weeks of quoting into hours, and gets sharper with every program.
Same engineering-drawing benchmark, precision and recall combined. Higher is better.
Team credentials and supported by
The quoting layer
that compounds with your team.
Drawings read at full physical depth
2D mechanical drawings, 3D CAD math, SolidWorks assemblies, and factory floor layouts parsed as engineering content, with dimensions, GD&T, tolerances, and surface finishes read directly from the drawing. Nyotta AI reaches 0.91 F1 on an engineering-drawing benchmark where general-purpose AI scores 0.38.
0.91 vs 0.38
F1, drawing benchmark
Every quote makes the next one smarter
Nyotta AI turns every drawing, quote, supplier price, and engineering decision into one living, searchable knowledge base, so your hard-won expertise stays in the business and gets sharper with every program.
Living context
Per customer, per program
Engineer in the loop on every call
Specialized agents draft. Your team approves. Nothing writes to your CRM or ERP without engineer sign-off, and every number traces back to the exact drawing callout or spec line it came from.
Approve to write
Full audit trace
How Nyotta works.
Reads the full RFQ package
2D drawings, 3D CAD math, SolidWorks assemblies, mixed-format spec packs, and factory floor layouts ingested in one pass. Component interconnections traced across the layout so spec compatibility is checked at the system level.
Specialized agents cross-validate
A fleet of specialists, each tuned to its job: extraction, technical reasoning, estimation, and compliance. Cross-checked against your historical programs, supplier data, and live market signals at the line-item level.
Engineer approves, system updates
Every recommendation surfaces its full reasoning trace. Your team reviews, modifies, or approves. Nothing writes back to your ERP or CRM without explicit human approval.
Built for engineering depth
generic AI cannot reach.
Generic LLMs
- Shallow API-level access to your systems
- Cannot reliably extract from 2D drawings, 3D CAD math, or mixed-format spec packs
- No awareness of your historical programs or supplier relationships
- Single-model, single-pass with no specialist depth
Nyotta AI
- Deep semantic parsing across drawings, CAD, floor layouts, and mixed-format documents
- Variant-resolved BOMs across thousands of configurations per program
- Continuous context: live commodity prices, supply chain shifts, historical program intelligence
- A fleet of specialized agents with independent eval harnesses that compound with every engineer correction
Built by people who lived the problem firsthand.
Maddie Zheng
Co-founder & CEO
Led the Pixel A-series program at Google from concept to launch. Built AI infrastructure for large-scale models at Meta Superintelligence Labs, including trust and safety systems running inline at production scale. Dropped out of Harvard Business School to build Nyotta AI.
Yiyuan
Co-founder & CTO
Led NLP and AI projects at Google Assistant (now part of Google DeepMind). Built enterprise search at Moveworks for Northrop Grumman, CVS, and other enterprises with the most fragmented legacy systems and highest security requirements. Graduate degree in CS and NLP from Yale.
See it on your own RFQs,
end to end.
We work with your existing formats. No integration required to start.