CopperPilot Raises $390K Pre-Seed for PCB Design Agent
A software developer can now hand a circuit-board assignment to a specialist agent without leaving the coding environment. CopperPilot has built that handoff into its electronics-design business, extending the reach of an AI assistant into the files and decisions that eventually become physical hardware.
The Bengaluru and Houston company announced a $390K pre-seed round led by Neon Fund on October 27, with participation from US-based angel investors. The announcement accompanies the launch of Epsilon, its first-version design agent for hardware engineers, and funding intended for product research, engineering improvements and team expansion.
Co-founders Aditya N G, CEO, Anish Harsha, COO, and Mohit Bagur are building around engineers' existing electronic design automation tools. Their opportunity sits in the work between describing a product and reviewing an actual board: finding parts, building schematics, arranging components, routing connections and checking whether the design respects its constraints.
A design agent that stays inside the engineering workflow
CopperPilot's product works on printed circuit boards, the assemblies that connect electronic components inside a device. KiCad provides its clearest demonstrated workflow, while the company also reports support for Altium and LTspice. Keeping the existing design environment gives engineers a familiar place to examine what the agent produces and decide which changes to accept.
That choice reaches beyond the interface. An electronics project contains decisions about component behavior, physical placement, manufacturing rules and the relationship between the board and its firmware. Assistance becomes more useful when those decisions remain visible as the project moves through successive stages. A neat schematic can still leave the next engineer with difficult questions about the layout.
The company's September integration documentation describes a command-line client, portable skill, Claude Code plugin and Model Context Protocol server. These let coding-agent hosts delegate work to CopperPilot's own PCB agent. The documented arrangement preserves a specialist responsible for the board rather than asking a general coding assistant to make concurrent edits to an active design.
For a team building a connected device, that creates a route from software work into electronics work through a tool it may already use. The commercial significance is reduced friction at the handoff: an engineer can reach the specialist without moving the entire project into a replacement design platform. Integration establishes that route; it does not establish a commercial partnership with the host-tool vendors.
What the financing supports
CopperPilot was founded in October 2025, according to the funding announcement. Aditya N G has described hardware projects, including autonomous vehicles, that exposed the time consumed by datasheets and repeated board revisions. The company is now asking engineers to bring that kind of work to an agent while retaining their ability to inspect the output.
Neon Fund invests at the pre-seed and seed stages in Enterprise AI. Leading this round places its capital behind an application with a specific professional user and an established engineering workflow. CopperPilot says it plans to use the financing to improve the agent's speed, reliability and design capabilities, alongside expanding its engineering team.
The announcement provides an early activity signal. CopperPilot reports 25-fold growth in weekly active users since August 2026 and more than 6,000 design tasks run on the platform. The absolute weekly user count and revenue were not disclosed. Those measures describe engagement with the product, while the quality of a finished electronic design requires evidence closer to the board itself.
The financing also has a disclosure boundary worth keeping clear. An October 9 SEC filing recorded $300K offered and sold, with a first sale on September 11. The filing and the later announcement are dated disclosures whose relationship has not been explained publicly. Adding them would manufacture a lifetime-funding figure that neither source supplies.
The evidence lives in designs engineers can inspect
CopperPilot's October evaluation write-up describes 341 KiCad tasks drawn from 49 designs, with its specialist agent and a general-purpose agent using the same underlying language model. Grading relied on KiCad checks and scripted tests rather than an AI judge. The company also identifies routing on larger boards as an area where further work remains.
That method gives a prospective user something more useful than an isolated demonstration. It identifies the work being tested, the comparison and some of the conditions under which the results were produced. The evaluation remains company-run and based on a single evaluation seed, so its scores should be read within that setup rather than treated as a forecast of manufacturing yield.
The distinction matters because a circuit has several ways to be wrong. Electrical rule checks and design rule checks cover important constraints, but the engineer must also consider the behavior of components, signal relationships and the requirements of the product being built. In a direct EE Times interview, Aditya N G describes the agent's role in reviewing issues beyond baseline checks and producing reports that prioritize remaining problems.
For buyers, a practical evaluation can therefore follow a board through preparation, review and revision. The output should give the engineer enough information to understand the choices, challenge them and make the next decision. That is a purchasing question about professional work, even when the first interaction begins with a plain-language prompt.
Bringing the next revision back into the product
Public engineering feedback is already reaching CopperPilot's founders. In responses on his LinkedIn profile, Mohit Bagur acknowledged specific concerns about connector spacing, ground reference and signal separation on an example design. His replies described changes for a subsequent revision and the importance of an experienced reviewer's judgment.
That exchange makes the company's operating task unusually tangible. An assistant preparing a layout has to preserve constraints that matter to both the electrical system and the person handling it. Feedback about the distance between connectors belongs in that work as surely as a component specification does.
CopperPilot's new financing supports a team working to bring those details into an iterative design process. As Epsilon reaches more engineers, the useful conversation continues in the project files: which requirement survived the last change, which issue has been fixed and whether the next board is ready to leave the screen.
Frequently Asked Questions
How does CopperPilot fit into an existing PCB design workflow?
CopperPilot assists with component research, schematics, layout, routing and review inside existing engineering tools. KiCad is its clearest demonstrated workflow; the company also reports Altium and LTspice support.
What does the Epsilon integration with coding agents do?
CopperPilot documents a CLI, skill, plugin and MCP server through which hosts such as Claude Code, Cursor and Codex can delegate board work to its specialist agent. The integration is a technical handoff, not evidence of a commercial partnership with those vendors.
How should engineers interpret CopperPilot’s evaluation results?
The company describes 341 KiCad tasks across 49 designs, with engineering-tool and scripted grading. It is a company-run evaluation using one evaluation seed, and it identifies large-board routing as an area still needing work.
What is the new funding intended to improve?
CopperPilot says the $390K pre-seed will support product research, engineering improvements in speed and reliability, and engineering-team expansion. These are stated plans rather than completed outcomes.
Can the earlier SEC filing be added to the funding announcement?
The October 9 filing recorded $300K sold, while the October 27 announcement describes $390K pre-seed. Their relationship has not been explained publicly, so adding them would produce an unsupported total.
Where the Money Moved
The intelligence briefing of the innovation economy. Funding, M&A, debt and fund closes, read as market signal rather than deal announcements.
Subscribe to Where the Money Moved





