Deeptech startup Maieutic Semiconductor raises $4.15 million from Endiya Partners, Exfinity Venture Partners – The Economic Times
The Bengaluru-based startup was founded by Gireesh Rajendran, Ashish Lachhwani, Rakesh Kumar, and Krishna Sankar. Maieutic is developing what they call the world’s first GenAI copilot for analogue design. This platform aims to speed up the early stages of chip development, automatically find bugs, and improve decision-making around design trade-offs.
With the fresh funds, Maieutic plans to expand its engineering team and significantly improve time to market. The company is also hiring to build out its platform. The deployment of its product is yet to begin.
Semiconductor design has resisted change and modern productivity enhancements, cofounder and chief executive officer Rajendra told ET. “Maieutic’s copilot can reduce the design cycle from weeks to days, spot inconsistencies without expert intervention, and bring intelligence to every trade-off,” he said.
The company set out to create its own AI platform because chip design requires a lot of domain-specific know-how, which is absent in a generic model. So, the first task was to build a clean enough data set that could be used to aid specific circuit designers.
“When we go through the process, there are lots of manual efforts related specifically to circuit design in creating test benches, drawing circuits, connecting outputs, and probing. So, with this agentic workflow, there is room to automate all these non-creative tasks, which leaves the designer to focus only on the creative tasks,” CTO Sankar explained.
Hallucinations in AI is a crucial aspect to solve for, Sankar added, because for the circuit designer, accuracy is key. “The tool will have enough guardrails or training data around it to help make sure that the designer gets the accurate responses,” he said.
“Maieutic is solving a real problem in the semiconductor design space, an area that has long resisted automation despite its growing complexity. Analogue workflows in particular have remained largely manual and dependent on domain expertise and time-intensive iteration,” Sateesh Andra, managing partner at Endiya Partners, said in a statement.