Summary
Abdullah Mohamed, VP of AI/ML at AIDAChip, presents at AI Engineer on a problem that rarely gets discussed in AI conferences: the organizational alignment bottleneck inside hardware engineering teams. The core thesis, backed by interviews with 15 chip design practitioners, is that engineers spend approximately 70% of their time on alignment — ensuring that specs, design decisions, tool outputs, and stakeholder approvals are consistent across a team — rather than on the actual engineering. With the average chip re-spin costing around $50 million and some companies facing make-or-break consequences from a single month’s delay, the stakes for getting this right are far higher than in software.
AIDAChip’s solution is a multi-layer AI system organized around what they call a System of Intent — a living knowledge graph that continuously ingests design decisions, Slack messages, wiki entries, tool inputs and outputs, and simulation results, replacing the fragmented state that normally lives across disconnected systems. On top of this graph, engineers get role-based AI teammates: specialized agents for digital design, analog design, and other sub-disciplines, trained by subject matter experts rather than general-purpose coding agents.
The demo shown includes three flows: a human engineer signing off on a simulation result that automatically notifies downstream stakeholders; the system detecting an out-of-constraint value in the graph and alerting the relevant engineer before it propagates to silicon; and a proposed design change that fans out through the graph to collect all affected stakeholders, gets approved by an architect, and then propagates the update to every dependent component in the system. The evaluation philosophy is based on compounding context quality over time rather than single-session accuracy metrics.
📺 Source: AI Engineer · Published August 22, 2026
🏷️ Format: Keynote Launch
