Design Time
About 73% less design time on medium-complexity boards
Roughly 40 hours of conventional design work reduced to about 10.8 hours for boards with 51 to 200 components.
Validation
Benchmark Results, with the Method Attached.
These figures come from internal benchmarking of the @I Design pipeline against conventional EDA workflows. They show what the system achieved under test conditions, not what every project will achieve.
Each result is published with its test scope and limits, so you can judge how far it applies to your own boards.
- Source: internal benchmark documented in NZ patent application 820555.
- Sample: 500 design scenarios derived from industrial projects.
- Benchmark results, not guarantees for any individual project.
Benchmark Results
Measured Against Conventional Workflows
Each design scenario was completed twice: once with conventional EDA workflows and once with the @I Design pipeline.
First-Pass DRC
First-pass DRC pass rate up from 65% to 92%
The share of designs that cleared design-rule checks on the first attempt. A DRC pass is a necessary step towards fabrication, not a measure of manufacturing yield.
Trace Length
18% shorter total trace length on average
Average reduction in total routed trace length across the benchmark set.
Via Count
24% fewer vias on average
Average reduction in via count across the benchmark set.
Test Method
How the Benchmark Was Run
The benchmark was built to compare like with like.
- 500 scenarios derived from industrial projects dated 2015 to 2023: 150 simple (1 to 50 components), 200 medium (51 to 200) and 150 high-complexity (201 or more).
- Domains included consumer electronics, industrial automation, automotive, medical, telecommunications and aerospace.
- Paired runs with equal time and computing budgets, and requirements supplied in the same format to every workflow.
- Quality assessed by expert reviewers who did not know which workflow produced each design; each workflow was operated by five designers of varying experience.
Evidence Boundary
How to Read These Results
Benchmark results describe system behaviour on a defined test set. They are not a promise about any particular board.
- These are internal results, not independent third-party measurements.
- Outcomes depend on board complexity, the quality of the brief and constraints, and review cadence.
- Within a program, progress is measured against the acceptance criteria agreed at intake.
Validation Summary
Published Figures, Stated Limits
A figure is only useful with its conditions attached. Results relevant to your board class and constraints are discussed in an architecture session.