Read run evidence.
Timing, congestion, DRC, constraints, logs, and floorplan context land in one view.
ArchGen AI
Self-learning agents for chip design.
ArchGen is building Newton, an autonomous co-engineer for backend physical design.
Mission
Ideas to precise hardware. Faster and smarter.
A future where hardware are no longer the bottleneck to human progress.
Learning loop
Autonomous PD loops, not manual triage.
Newton reads run evidence, points to root causes, adjusts flows, and carries the outcome into the next iteration.
Find root causes.
The agent maps symptoms to likely issues across placement, routing, constraints, and flow settings.
Iterate the flow.
Newton proposes floorplan or script changes and prepares the next run with clear reasoning.
Keep the outcome.
Accepted fixes, rejected paths, and final results become memory for the next block.
DEMO
How We Ranked First in the HRT (Hudson River Trading) and Partcl Macro Placement Challenge
A technical write-up of the approach behind our verified rank-1 score on the IBM macro-placement benchmarks.
Why Chip Design Still Takes 18 Months, and What AI Can Actually Fix
Where time disappears, and what better memory could change.
Cognitive Agents for the Next Era of Silicon
A practical look at context, feedback, and workflow memory.
On Disrupting the Semiconductor Industry with AI Agents
Less magic. More context. Better loops around real engineering work.
BLOGS
FAQS
What is ArchGen AI for semiconductor physical design?
ArchGen AI is building Newton, a self-learning agent for semiconductor backend physical design. It helps teams use run evidence, fixes, and outcomes from prior work when they face the next timing, congestion, DRC, or signoff problem.
How do AI agents help physical design teams?
When a PD run hits timing, congestion, or DRC issues, Newton reads the reports, identifies likely root causes, and helps iterate on floorplans and flows.
What does design memory mean in chip implementation?
It means preserving useful backend engineering context across runs, including constraints, timing paths, congestion issues, DRCs, ECOs, fixes, and outcomes.
Is ArchGen replacing EDA tools?
No. Newton is being built as an autonomous backend engineering layer around physical design and EDA flows, not as a replacement for existing tools.
How can AI help with timing, congestion, and signoff?
It can read reports, find likely root causes, propose flow or floorplan changes, and reduce repeated manual triage and rerun cycles.
Who is Newton for?
Newton is for physical design and implementation teams working through backend runs, floorplans, timing closure, congestion, DRCs, ECOs, and signoff loops.
What evidence does Newton learn from?
Newton is designed to learn from run reports, constraints, timing paths, congestion maps, DRCs, ECO decisions, scripts, fixes tried, and final outcomes.
Is this just another automation script?
No. Scripts repeat known steps. Newton is being built to preserve context, reason over prior outcomes, and suggest the next useful change when a design run hits a familiar problem.
Can't find what you're looking for? Reach out to us at hari@archgen.tech.
Connect
Working on backend physical design?
Subscribe for notes or reach out if you are thinking about self-learning agents, PD automation, timing closure, congestion, DRC, or signoff.