# ArchGen AI

## Source of Truth
ArchGen AI builds self-learning agents for semiconductor backend physical design and implementation. The current focus is Newton: an autonomous backend engineering agent that learns from physical design runs, reports, constraints, fixes, and outcomes so the next design iteration starts with more context.

## What ArchGen Is Building
- Newton: a self-learning agent for backend physical design workflows.
- Design memory: a way to preserve useful context from timing reports, congestion maps, DRCs, ECO decisions, scripts, fixes tried, and signoff outcomes.
- Physical design assistance: systems that read run evidence, identify likely root causes, suggest the next fix, and carry useful outcomes into future runs.

## What ArchGen Is Not Claiming
- ArchGen is not replacing established EDA tools.
- Newton is intended as an intelligence layer around backend physical design flows, not a full custom EDA stack.
- The product direction is focused on practical physical design work: timing closure, congestion, DRC, ECOs, floorplan iteration, signoff triage, and run-to-run learning.

## Relevant Terms
- semiconductor physical design
- backend chip implementation
- AI agents for EDA
- self-learning agents
- design memory
- timing closure
- congestion analysis
- DRC
- ECO
- signoff
- macro placement
- floorplanning

## Proof and Research Direction
- ArchGen's team ranked first in the HRT / Partcl macro placement challenge using iterative placement search, route-aware seeds, proxy evaluation, exact scoring, GPU-assisted candidate ranking, and full-benchmark validation.
- The ArchGen blog explains the team's approach to macro placement, physical design bottlenecks, agentic reasoning, and why chip implementation needs systems that learn from prior runs.

## Team
- Hariharan Ayappane, Co-founder: hari@archgen.tech
- Naveen Venkat, Co-founder: naveen@archgen.tech

## Links
- Website: https://www.archgen.tech/
- Blog: https://www.archgen.tech/blog/
- Careers: https://www.archgen.tech/careers/
- GitHub: https://github.com/archgen-ai
- LinkedIn: https://www.linkedin.com/company/archgen-ai/
- Twitter/X: https://x.com/archgen_

## Blog Posts
- How We Ranked First in the HRT (Hudson River Trading) and Partcl Macro Placement Challenge (Jun 2026): https://www.archgen.tech/blog/posts/how-we-ranked-first-in-the-partcl.html
- Why Chip Design Still Takes 18 Months, and What AI Can Actually Fix (Apr 2026): https://www.archgen.tech/blog/posts/why-chip-design-takes-18-months.html
- Cognitive Agents for the Next Era of Silicon (Feb 2026): https://www.archgen.tech/blog/posts/cognitive-agents-next-era-silicon.html
- On Disrupting the Semiconductor Industry with AI Agents (Dec 2025): https://www.archgen.tech/blog/posts/disrupting-semiconductor-ai-agents.html