Mach9 Built Hardware. Customers Said Hardware Was Not The Pain.

Picture a founder walking into a meeting with a shiny prototype. The prospect looks at it, nods politely, and says, "Cool. But that's not really what slows us down."

That's roughly what happened to Alex Baikovitz. He co-founded Mach9 in 2021 to give civil engineers a faster, cheaper way to do visual inspections. On October 21, 2026, TechCrunch reported that the company, now without "Robotics" in its name, raised a $34 million Series A for AI-powered CAD software.

Same founder. Same industry. Very different product. And the hinge was a sentence you can collect for free.

Mach9 moved from building inspection hardware to building software that turns scan data into a first-draft plan. According to TechCrunch, the team spent its early years on the hardware side. Then Baikovitz concluded that hardware wasn't the real constraint. Turning data from laser scanners and imaging systems into usable site surveys was the heavier burden.

His summary: "The real pain was in their workflows."

The company dropped "Robotics" from its name and went after that workflow. The Series A was co-led by First Round Capital and Bessemer Venture Partners, with Prudence and existing investor Quiet Capital also taking part.

Notice what didn't happen. Nobody told Baikovitz his robots were bad. They told him the robots weren't the problem.

Because praising a prototype is free, and telling you what actually hurts takes effort. Most people are nice. If you show them a gadget, they'll say it's neat. They won't volunteer that their real headache lives three steps downstream.

TechCrunch notes that the earlier hardware work still paid off. It gave the team something to show prospects, and those prospects said they didn't need cheaper devices. Read that again. The prototype wasn't the product. It was a conversation starter that exposed the real job.

I've seen this play out more than once with early teams: the demo is a door opener, not a destination. The founders who learn fastest treat every demo as a question. "If this worked perfectly, what would you still be stuck on?"

The slow, manual work between raw scan data and a finished drawing. Baikovitz told TechCrunch that delivering a project the manual way in incumbent tools from Autodesk and Bentley Systems can take months per project, with roughly a week of work per mile.

Mach9 says its software is about 10 times faster than manual work and substantially lowers costs. It also reports nearly 100 customers in a private beta and use on more than 2,500 projects. Those are company claims, so treat them as claims. Still, the pattern is useful: the buyer's pain was hours of skilled labor, not the price of a sensor.

Here's the question to steal. In your customer's week, where does a skilled, expensive person sit and do something tedious? That's where budgets already exist.

Not quite, and that's the interesting bit. By November 2024, when Mach9 announced a $12 million seed round led by Quiet Capital, Geo Week News described it as a software company turning mobile LiDAR scans into digital maps. Its product, Digital Surveyor, detected features such as utility poles, traffic signals, and road signs. At the time, the company claimed mapping up to 96 times faster than traditional methods.

TechCrunch says the team has focused on training its own AI models since 2024 to reach the centimeter-level accuracy surveying work requires.

So the "pivot" wasn't a single dramatic Tuesday. It was a slide. Hardware first, then mapping software, then a broader CAD product for engineers. Y Combinator's profile, which lists the company as part of its Summer 2021 batch, now calls it "AI CAD Software for Civil Engineering."

Big pivots usually look sudden from outside and gradual from inside.

You don't need a funded round. You need ten conversations and a notebook. Here's a rough script.

Ask the person to walk you through their last project, step by step. Don't pitch. Note where they sigh, where they say "well, it depends," and where they mention waiting. Ask how long each step takes and who does it. Ask what they've tried to fix it, and what that cost.

Then show your idea, and ask the Mach9 question: "What would you still be stuck on?"

If three different people name the same step, you've found a candidate. If everyone shrugs, your idea might be a solution hunting for a problem. And if they love the idea but won't change how they work, that's a warning, not a win.

Once you have your answers, you have to put them somewhere. You can sort the notes in a spreadsheet, a Notion board, or a planning tool like Foundra, which walks first-time founders through sections such as customer, problem, and go-to-market. Any of them works if you actually revisit the page.

What matters is the structure. Write the problem as the customer said it, in their words. Next to it, write the evidence: who said it, when, and how many times. Then add a column for "what I assumed before the call."

When the right column and the left column disagree, you've just found your pivot. It costs nothing but an afternoon.

Be careful about the wrong lesson. A big round isn't proof you were right. It's proof that someone believes the new version is bigger than the old one.

What the Mach9 numbers do suggest is that investors like a founder who can say, specifically, why they changed course. "Nearly 100 customers in beta and 2,500 projects" is a story about demand arriving after the change. If you can't tell a story like that, a pivot just looks like drifting.

Also, TechCrunch placed Mach9 among other companies chasing AI for physical-world engineering. CivilGrid raised $26 million in August for what it calls a "Google Maps for the underground," and Blueprints AI, which generates permit-ready architectural drawings, reached the top 200 in Startup Battlefield at TechCrunch Disrupt. That's a crowd. A real customer pain narrows your field faster than a clever demo does.

Pick one customer type. Book five calls. Ask for a walk-through of the last project, not an opinion about your idea. Write down every step that involves a skilled person waiting, redoing, or retyping.

Then ask yourself an uncomfortable question: if I removed the part of my product I'm proudest of, would anyone notice?

Mach9 had the nerve to ask that and to drop a word from its own name. You don't need to rename anything. You do need to find out which part of your plan the customer would skip.

Sometimes they can't, because they've lived with it so long it feels normal. A surveyor who has spent ten years redrawing the same road probably doesn't think of that as a "problem." It's just Tuesday.

So watch, don't only ask. If you can sit in on a real project, even for an hour, you'll spot the workarounds: the spreadsheet taped to the process, the colleague who always gets the annoying task, the file named final_v7_REAL.

Workarounds are gold. Every one is a customer building their own bad version of the product you could build properly. Count them, photograph them if allowed, and bring them back to your team.

Then price the workaround. If someone burns eight hours a week on it at a loaded cost of 60 dollars an hour, that's close to 25,000 dollars a year of pain sitting there. Your software doesn't need to be magic. It needs to cost less than the pain.

Demos are questions, not answers. The best pivot signal is often a polite sentence like "that's not really our problem." Look for skilled people doing tedious work, because that's where budgets live. Write the customer's words next to your assumptions. And remember that Mach9's change took years and several steps, not a weekend.

If your prototype gets praise but no pull, treat that as data.

What did Mach9 announce on October 21, 2026? TechCrunch reported a $34 million Series A co-led by First Round Capital and Bessemer Venture Partners, with Prudence and Quiet Capital also participating.

Why did Mach9 pivot away from hardware? CEO Alex Baikovitz told TechCrunch that hardware wasn't the real constraint. Converting scan data into usable site surveys was the bigger burden, and prospects said they didn't need cheaper devices.

What does Mach9's software do? It uses in-house AI models to turn survey data into a first-draft plan that engineers refine. Licensed professionals stay responsible for the final deliverable.

How many customers does Mach9 have? TechCrunch reports nearly 100 customers in a private beta and use on more than 2,500 projects. These figures come from the company.

How can a first-time founder test whether their real problem is different from their product? Interview ten target customers about their last project, ask where time goes, then show your idea and ask what they'd still be stuck on.

Is a pivot a sign the first idea failed? Not necessarily. Mach9's hardware work gave the team something to show prospects, and those conversations pointed to the workflow problem.

1. Mach9 raises $34M after pivoting to AI CAD software, TechCrunch, October 21, 2026
2. Mach9 Raises $12 Million Led by Quiet Capital to Scale AI-Powered Geospatial Mapping Solutions, Geo Week News, November 2024
3. Mach9: AI CAD Software for Civil Engineering, Y Combinator company profile
4. Mach9 Raises 12.0M USD in Seed Funding, Seedtable