Squint Gives Factory Workers AR-Guided Machine Instructions via Phone

01 /The Concept

Point a phone camera at a piece of industrial equipment and get step-by-step video instructions for operating it. That is Squint's core premise – and it turns out to be worth $19 million to Sequoia and others.

The platform serves manufacturing workers who need to operate and maintain complex machinery. The augmented reality layer is the differentiator: rather than navigating a manual or searching a knowledge base, a worker simply aims their phone's camera at a machine. Squint identifies it visually, pulls up associated video procedures, and can overlay directional annotations on screen – pointing to the specific control that needs to be adjusted and explaining what to do with it.

Creating content for the platform requires no specialist tooling. A factory expert performs a procedure while someone records it; the video gets uploaded; the AI parses it into a sequenced list of steps with text annotations. The same processing builds a controls database for each machine – cataloging every button, valve, and indicator with its associated actions – which feeds the in-app assistant that answers operational questions in plain language.

Scheduled maintenance runs through the same system: each machine can have a defined maintenance checklist with video instructions attached, reminder notifications pushed to the right workers at the right times, and a completion log visible to supervisors. Workers can also attach notes to specific machines or components – passing information between shifts, flagging suspected faults for service teams, or preserving observations that would otherwise disappear when a shift ends.

Client logos include Hershey, Volvo, Siemens, Colgate-Palmolive, and Michelin. The company raised $2.5 million in March 2022, $3.5 million in April 2023, and $13 million now – with Sequoia leading both recent rounds.

02 /Why It Matters

The market for this kind of platform is larger than most people assume. There are roughly 700 million manufacturing workers in the world – a number that dwarfs the global developer population of under 30 million. The amount of investment that has flowed into developer education tools versus manufacturing worker enablement is almost comically disproportionate.

The structural case for platforms like Squint is getting sharper. Manufacturing equipment has become genuinely complex, and the workforce that knows how to operate it is thinning. Industry turnover in manufacturing runs close to 40% annually – nearly half the workforce cycles out each year and must be replaced and trained. A Deloitte analysis found it nearly 50% harder to hire qualified manufacturing workers in 2020 than it was in 2018. That gap has not closed.

The economic stakes are direct. An equipment failure that stops a production line is not an inconvenience – it is a measurable loss per hour, compounded by repair costs and replacement parts. The case for preventing those failures through better-trained workers and faster expert access is quantifiable in ways that make procurement conversations straightforward.

DeepHow, [covered previously](/review/63-milliarda-dollarov-na-700-millionov-uchenikov), uses similar video-to-instruction processing and has raised $37.1 million. Zaptic, [reviewed here](/review/programmirovat-nuzhno-ne-kompjutery-a-ljudej), takes a comparable on-floor guidance approach and has raised $16 million. Kognitiv Spark adds live expert AR assistance for real-time troubleshooting. Each company has a different angle on the same core problem – which is itself evidence that the problem is large enough to support multiple approaches.

03 /Opportunities

The market size is the simplest argument here. Seven hundred million workers is large enough that a dozen well-funded platforms could each capture a meaningful slice without meaningfully competing. The constraint is not demand – it is coverage. Most manufacturing facilities that could use a platform like Squint have not yet heard of it, evaluated it, or budgeted for it.

Squint's combination of low-friction content creation (video upload plus AI parsing) and AR-guided delivery is a reasonable model to study. The hardest parts of building in this space are not technical – they are the sales motion (enterprise manufacturers are not fast buyers) and content seeding (a platform with no video library has limited immediate value to new customers). Both are solvable with the right go-to-market approach, and the incumbents have demonstrated that industrial clients will pay for solutions that demonstrably reduce downtime and training costs. The market has proven itself; the question is execution.