Epsilon Goes to Work: MyTechDailyNews Explains What CopperPilot's $390,000 Pre-seed Means for Hardware Teams

Most readers never look at the printed circuit board inside their router, their car's control unit, or the battery pack of an e-bike — and that invisibility is exactly why MyTechDailyNews thinks a small, early-stage hardware funding round is worth decoding. CopperPilot, a startup building an AI agent for circuit-board design, has closed a $390,000 pre-seed round led by Neon Fund, with participation from angel investors based in the U.S. The timing is deliberate: the raise lands as version 1 of the company's design agent, codenamed Epsilon, is opened to working hardware engineers. CopperPilot also points to 25-fold growth in weekly active users since August 2026, and more than 6,000 design tasks — artifacts, in its own vocabulary — already run on the platform.

The headline facts, laid out simply

• The money: $390,000 in a pre-seed round led by Neon Fund, joined by U.S. angel investors.
• The product: Epsilon, an AI design agent for PCB work, now reachable by engineers at version 1.
• Traction claimed: 25-fold growth in weekly active users since August 2026, plus more than 6,000 design tasks processed to date.
• Company: founded in October 2025 by Aditya NG, Mohit Bagur, and Anish Harsha, running dual headquarters in Bengaluru, India and Houston, Texas.
• Not disclosed: revenue, and the share of free users who convert into paid accounts.
• Use of funds: product research and engineering, better agent speed and reliability, and growth of the engineering team.

Why a PCB agent earns a place in a general tech feed

Hardware is where AI promises collide with physics. A wrong assumption in software becomes a patch; a wrong trace on a board becomes a new prototype run, a new supplier conversation, and a delayed launch. CopperPilot's pitch is that an agent can absorb the repetitive middle of that process — interpreting requirements, shortlisting components, drafting schematics, routing boards, and checking designs before manufacturing — while leaving judgment calls to humans. The company's stated goal is to ensure that “no hardware team is ever blocked because a board is too hard to design.”

That scope reaches past engineering departments. Consumer devices, medical equipment, energy storage, and robotics all inherit schedule risk from board design, so any tool that compresses the design loop eventually shows up in products ordinary readers buy.

The founders, in their own context

From left: Mohit Bagur, Anish Harsha, and Aditya NG, the co-founder team behind CopperPilot.

In an exclusive interview with EE Times, co-founder and CEO Aditya NG traced the idea to hardware projects he had worked on, including autonomous vehicles. “On almost every autonomous vehicle team I worked with, someone would eventually say it: A compute unit for six cameras is too hard to design, or a drive-by-wire unit is too much for us. Many teams stopped right there. The ones that kept going spent months staring at datasheets and paying for mistakes on board re-spins,” he said. “I built CopperPilot so that no hardware team is ever blocked because a board is too hard to design,” he said.

Adoption strategy: live inside the tools teams already use

Rather than asking engineers to migrate designs into a new environment, CopperPilot automates workflows inside existing EDA tools. KiCad, Altium, and LTspice are supported, with the deepest integration around PCB design and firmware work. Epsilon can also be reached through AI coding agents — Claude Code, Cursor, and Codex — so a developer can hand board-design tasks off without leaving that context. To interact with EDA tools, the company blends official application programming interfaces (APIs), direct file editing, and other techniques. “The method varies according to what works for a particular tool or task,” Aditya said.

Tool or environment	Where CopperPilot says it plugs in	Adoption implication
KiCad	PCB design workflows; the team's open-source board designs can be inspected here	Low-friction trial for hobbyists and small teams
Altium	Schematic and board work in a professional PCB flow	Fits teams with established libraries and processes
LTspice	Simulation-adjacent checks around circuit behavior	Keeps analysis close to design instead of in a separate silo
Claude Code, Cursor, Codex	Board tasks handed off from inside an AI coding agent session	Developers and firmware engineers can delegate hardware work
GitHub and ticketing systems	Organizational context and shared knowledge for enterprise use	Company history, not only prompts, informs the output

From a plain-English sentence to a checked layout

Aditya described the workflow as a chain of small, inspectable steps:

1. Epsilon takes design requirements written in plain English.
2. It reads component datasheets and manufacturer specifications.
3. It identifies suitable components and presents alternatives for the engineer to choose from.
4. It creates schematics, plus custom symbols or footprints where needed.
5. It moves into PCB placement and routing.
6. It checks for electrical rule check (ERC) and design rule check (DRC) violations and for manufacturing constraints.
7. It asks clarifying questions when requirements are incomplete or components do not meet design constraints — the architectural decision layer.
8. When requirements are sufficiently clear, it executes a sequence of design tasks autonomously, while engineers can still review or intervene at individual stages.

Cross-tool work is further along the roadmap. In conventional practice, an engineer may inspect firmware code, PCB layout, and communication protocols separately. By connecting those sources, CopperPilot argues it can reason about how parts of an electronic system affect one another, not merely whether one board is internally tidy.

Validation is where the difficulty concentrates

Drawing a layout is only half the job. The design must also satisfy electrical requirements and the constraints of the manufacturer that will build it, and according to Aditya, ERC and DRC checks form a baseline rather than a complete answer.

He offered an example that many engineers recognize: “While connecting two computers or two microcontrollers over a serial connection, you are supposed to connect the transmitter of one device to the receiver of the other, and the receiver of the first device to the transmitter of the other,” he explained. “It is very easy to reverse these connections, and it happens to many engineers. This is where CopperPilot really shines. It can catch mistakes that ERC and DRC checks cannot detect.”

Severity matters as much as detection. “For engineers preparing a design for manufacturing, the agent can produce a review report that separates blockers from non-blocking issues, prioritizes fixes, and explains the remaining problems,” Aditya said. “It may also identify design changes that could reduce cost.”

Because hardware design has no single equivalent of a software compiler or linter to certify an entire design, the startup built its own test rig. “With CopperPilot, we have internally built out our own harness, benchmarks, and tooling to validate the design,” he said. “The agent can read the datasheet, manufacturer specifications, and design tools, with the startup testing how well this process works through its harness and multi-agent framework.” Manufacturer-specific test suites, maintained component information, and a growing record of design failures feed that infrastructure.

How to read the performance numbers

Figures shared by Aditya describe an evaluation against approximately 49 test cases based on human-designed circuits. On that set, he said, the startup reported a task completion rate of 18% for its baseline system and 78% for CopperPilot. Two caveats change how useful that number is: the baseline system was not specified, and the results come from a small, company-run benchmark that has not been independently verified.

A more checkable artifact is the open-source program. The company says it publishes a new open-source board design each day, with around 30 designs published so far, each inspectable in KiCad — a way for engineers to judge real output rather than rely only on product demonstrations.

Plans and packaging

CopperPilot operates as a B2B software-as-a-service (SaaS) company, with tiers for individuals, teams, and enterprises.

Tier	Who it targets	What the company says about it
Starter	Individuals evaluating the tool	Free plan in the individual lineup
Pro	Individual engineers with regular workloads	Paid tier
Pro Plus	Heavier individual use	Paid tier
Ultra	Power users needing maximum capacity	Highest individual plan named
Team plans	Design service providers and internal groups	Paid, with shared skills and standards
Enterprise	Larger businesses	Contact the company about specific requirements and internal deployments

Enterprise positioning leans on integration and institutional memory. Connections with GitHub and ticketing systems let the agent draw on an organization's context; an engineer can create a reusable skill that checks designs against company manufacturing standards and share it with colleagues, folding accumulated engineering experience into the agent's workflow.

Common misconceptions

• “This replaces the PCB designer.” The described model keeps engineers in control at individual stages and adds review reports; the claimed gain is speed and fewer missed errors, not a headcount swap.
• “It is just a wrapper around rule checks.” ERC and DRC are positioned as the baseline, with the differentiated work being mistakes those checks miss, plus prioritization of what to fix first.
• “You have to move your designs into a new CAD tool.” CopperPilot says it works inside KiCad, Altium, and LTspice, and can be reached from Claude Code, Cursor, and Codex.
• “The 78% figure is an independent verdict.” It comes from a company-run benchmark of approximately 49 test cases with an unspecified baseline, not verified by a third party.
• “A pre-seed raise means the product is finished.” Reliability in day-to-day engineering use is described as the immediate challenge; on-premises hosting, for instance, is offered but has not yet been deployed for a customer.

FAQ

What did CopperPilot raise, and who backed it?

A $390,000 pre-seed round led by Neon Fund, joined by angel investors based in the U.S. The company says the new funding will support product research and engineering, improve the agent's speed and reliability, and expand the engineering team. No revenue figure or user-to-paid-customer conversion rate was disclosed.

What is Epsilon, in one sentence?

It is CopperPilot's AI design agent for PCB work — version 1 is now opening to engineers — that interprets requirements, selects components, draws schematics, routes boards, and checks designs before manufacturing.

How fast is it, according to the company?

“Today, CopperPilot designs an AI camera board in about 30 minutes,” Aditya said. Read that as a vendor claim about drafting speed, not a guarantee that a production-ready board emerges in the same window.

Does my design data leave my machine?

For standard individual and paid plans, the company says design data and chat history are stored locally on the user's computer, while information is sent to servers and a large language model (LLM) for processing. “CopperPilot has a zero data retention policy after an interaction,” Aditya said at the time of the interview. “We also offer on-premises hosting as an option, although it has not yet been deployed for a customer.”

Who is already using it?

Individual engineers, startups, design service providers, and larger businesses. IoT, health and medical technology, battery technology, electric vehicles, and other automotive applications are among its customers and target markets. Customers are located in the U.S., U.K., and Europe, while users are also present in India and other countries; consumer electronics, renewables, energy storage, and robotics are named as further candidates.

Action advice

1. Hardware leads: before piloting, name the workflow that hurts most — component selection, schematics, routing, or pre-manufacturing checks — so evaluation targets a real bottleneck instead of novelty.
2. Teams shortlisting the tool: start with the published open-source boards, around 30 designs so far, opened in KiCad; judge output quality before wiring anything into your own libraries.
3. Security and compliance reviewers: ask exactly which data leaves the machine for LLM processing, get the zero data retention policy claim in writing, and confirm whether on-premises hosting is contractually available.
4. Founders and product managers: treat the benchmark numbers (18% baseline versus 78% on approximately 49 test cases, company-run and unverified) as an internal signal, not an industry comparison.
5. General readers: watch whether reliability, rather than speed, improves over the next release cycle — that is the condition for everyday engineering adoption.

What to watch next

CopperPilot's near-term test is making the agent dependable enough that engineers use it in day-to-day work, while adoption spreads from individuals to startups and larger design organizations. The longer aim is shortening the path from an idea to a manufacturable electronic product.

Drawing on his own history designing electronics for a student race car, Aditya closed with a prediction that says a lot about how the founders want to be judged: “The racecars of the future will be designed with CopperPilot.”

Bottom line

A $390,000 pre-seed round is not, by itself, proof that AI has solved hardware design. What makes this announcement worth a general-interest look is the combination of a product now at version 1, concrete usage claims — 25-fold growth in weekly active users since August 2026 and more than 6,000 design tasks run — and a candid admission that validation and reliability are the hard parts. MyTechDailyNews will keep tracking how the team's dual headquarters in Bengaluru and Houston ship against that bar, and we will flag any independent benchmarking as it appears.

By Sarah Chen, MyTechDailyNews · Tech Daily newsroom. Story tips: hello@techdaily.com