Normal Wants to Make Hardware Certification as Simple as Booking a Flight

The Macro: Hardware Certification Is the Bottleneck Nobody Talks About

I want to talk about a problem that is invisible to almost everyone outside of hardware engineering but quietly kills products, delays launches, and burns through cash at an alarming rate. If you are building any physical product that uses wireless connectivity, emits electromagnetic radiation, or interacts with the human body, you need to get it certified before you can legally sell it. FCC in the United States. CE marking in Europe. ISED in Canada. ASTM for safety standards. And a dozen more depending on where you are selling and what the product does.

The certification process looks something like this. You design your product. You build a prototype. Then you need to find an accredited testing lab, book time on their equipment, ship your device, wait for test results, discover that your device fails on conducted emissions at 150 kHz, redesign the power supply, build a new prototype, ship it back, test again, wait for results, pass this time, compile a 200-page technical file, submit it to the regulatory body, wait for review, respond to questions, and finally receive your certificate. That process takes three to twelve months and costs anywhere from $10,000 to $250,000 depending on the complexity of your device.

For startups building drones, robots, medical devices, AR headsets, or autonomous vehicles, this is not a minor inconvenience. It is a fundamental business risk. You cannot sell your product until you have certification. You cannot raise your next round if your timeline to market depends on a testing lab that has a four-month backlog. Investors ask “when can you ship?” and the honest answer is often “whenever Bureau Veritas or UL Solutions finishes testing our device.”

The testing and certification market itself is dominated by a handful of massive companies. TUV Rheinland, SGS, Intertek, Bureau Veritas, and UL Solutions collectively control most of the accredited testing capacity worldwide. These are not technology companies. They are industrial conglomerates that move at industrial speed. Their customer portals look like they were built in 2008 because they were. Their communication happens via email chains and PDF attachments. Their pricing is opaque.

This is a market that is growing rapidly because the number of connected devices is exploding. Every new hardware category, from delivery robots to brain-computer interfaces, needs certification. And the infrastructure for processing those certifications has not scaled to match demand.

The Micro: Two Founders Streamlining the Path From Prototype to Certified Product

Normal is a platform that automates hardware certification. You describe your product, and the system guides you through the entire process from initial compliance assessment through lab booking, results review, technical file preparation, and final certification. The company claims it reduces certification timelines by 50x compared to the traditional process.

Anson Yu is CEO and Hudhayfa Nazoordeen is co-founder. They are based in San Francisco, part of Y Combinator’s Summer 2025 batch, with a two-person team. The tagline on their site is “Take the pain out of hardware certification” and the framing is deliberately cheeky: “Flying Cars? Normal. Surgical Robots? Normal. Compliance headaches? Abnormal.”

The 50x speed claim is the kind of number that needs context. If the traditional process takes six months and Normal cuts it to three days, that is impressive but probably only applies to simple consumer electronics with straightforward FCC requirements. A medical robot going through FDA and CE certification is a fundamentally different complexity level, and I would be surprised if any platform can compress that by 50x. What I think is realistic is that Normal eliminates the dead time, the weeks spent waiting for lab availability, the back-and-forth on technical documentation, and the confusion about which standards apply to your specific device.

The platform supports FCC, ISED, CE, and ASTM certifications. That covers the major markets for most hardware startups. The pre-compliance software component is interesting because the most expensive part of certification is failing a test. If you can run pre-compliance checks on your design before you ship a prototype to a lab, you catch problems early when they are cheap to fix instead of late when they require a hardware revision.

The lab booking feature connects hardware teams with ISO 17025 and ISO 17065 accredited testing facilities. This is essentially a marketplace model layered on top of the existing lab infrastructure. Normal is not building testing labs. They are making the existing labs more accessible and reducing the friction of finding one with availability, getting a quote, and booking time.

No pricing is listed on the site. The model is consultation-first, which makes sense for a product where the scope of work varies dramatically between customers. Certifying a Bluetooth speaker is fundamentally different from certifying an autonomous drone, and flat pricing would either underprice complex jobs or overprice simple ones.

The Verdict

Normal is solving a problem that will only get more painful as the hardware ecosystem expands. Every new category of connected device, every new market a hardware startup wants to enter, creates more certification work. The companies that control the testing infrastructure have no incentive to move faster because their backlogs are full regardless. A software layer that optimizes the workflow around those bottlenecks has clear value.

The question is how deep the product can go. If Normal is primarily a project management tool for certification workflows, the value is moderate. If it can actually accelerate pre-compliance testing and reduce the number of failed lab tests, the value is enormous. A single failed test can cost $20,000 in lab fees plus weeks of delay. Preventing one failure pays for the entire platform.

In thirty days, I want to know what product categories their early customers are certifying. That tells you whether they are starting with simple consumer electronics or tackling harder verticals early. In sixty days, the question is whether they can demonstrate measurable time savings with real customers. The 50x claim needs data behind it. In ninety days, I want to see whether the lab marketplace is creating network effects. If labs start preferring Normal-submitted applications because the documentation is more complete and standardized, the platform becomes sticky on both sides of the market. Hardware certification is one of those problems where the people suffering most are too busy to build a solution themselves. That is usually a good sign for the startup that builds it for them.