Squint has raised $40 million to expand a mobile platform that turns factory equipment and experienced workers' knowledge into visual instructions. The investment case rests on a difficult operational promise: making guidance easier to retrieve without allowing unverified advice to become the new standard.

On 15 August 2025, Forbes reported that the Series B valued Squint at roughly $265 million. The $40 million round was led by The Westly Group and TCV, with existing investors Sequoia Capital and Menlo Ventures participating. Squint said the financing brought its total raised to more than $59 million.

The product starts where a manual becomes inconvenient

Industrial troubleshooting often combines a machine-specific fault, a printed procedure and knowledge held by a small number of experienced technicians. Squint's mobile-first product uses artificial intelligence and augmented reality to connect the physical equipment seen through a phone with instructions, annotations and captured process knowledge.

That does not make the phone an autonomous machine controller. An operator still performs the work and remains inside the site's safety, lockout and authorization procedures. The useful distinction is between finding relevant guidance at the asset and allowing software to command the asset. The funding story concerns the first of those functions.

Gloved hands hold an unbranded phone toward a hydraulic component inside a CNC machine while simple geometric overlays match the digital view to the physical part
The interface is valuable when it shortens retrieval while preserving the plant's control over the job.

What a controlled instruction needs

  • a named asset, procedure owner and intended user role;
  • engineering and safety review before release;
  • a version number, effective date and change history;
  • clear prerequisites, isolation steps and stop conditions;
  • feedback from operators who encounter a different configuration;
  • evidence that the instruction improved the task without creating new risk.

The round finances scale rather than proving it

In its 12 August announcement, Squint said it would advance its AI capabilities, deepen implementation with manufacturers, meet demand and expand into energy, logistics and field services. It described tens of thousands of operators using the platform across hundreds of Fortune 500 factories.

Those are company statements, not an independently audited market census. Forbes named PepsiCo and Ford among users and described equipment ranging from CNC machines to forklifts. Customer names show industrial relevance, but they do not establish the scale, duration or outcome of every deployment.

The company is based in the United States, where manufacturers face the same knowledge-transfer problem as plants elsewhere: experienced workers retire, equipment variants accumulate and a generic document repository becomes hard to navigate during a fault.

Knowledge capture is a production process

Recording a veteran technician is only the input. The resulting sequence has to be decomposed into observable steps, matched to the correct machine revision and checked against engineering and safety requirements. Photos and annotations must avoid exposing personal or commercially sensitive information. Translations need technical review rather than literal substitution.

A mature workflow separates draft capture, validation, approval, release and withdrawal. When a component, tool or hazard changes, the instruction should be recalled or updated across every affected site. Otherwise digitization can distribute an obsolete shortcut faster than a paper binder ever could.

A worn industrial component moves through camera capture, modular instruction cards, engineering and safety approval gates before identical tool kits reach three factory stations
Scale depends on a governed content lifecycle, not simply on recording more procedures.

Performance claims need comparable baselines

Squint cited more than $4 million in increased profit at one unnamed Fortune 50 site, 50% faster procedure execution for first-time workers at an unnamed consumer-goods company and 91% operator satisfaction across thousands of sessions. These figures describe selected company-reported cases. They are not universal forecasts for prospective customers.

A buyer should define the baseline before rollout: time spent finding instructions, mean time to repair, first-time completion, scrap, rework, safety interventions and calls to expert technicians. Measures should be segmented by machine, task difficulty and worker experience. A faster easy procedure should not conceal worse outcomes on rare high-risk work.

A practical deployment sequence

  1. Choose one costly, frequent and well-bounded procedure rather than an entire plant.
  2. Confirm the approved source of truth and who can release changes.
  3. Test identification across lighting, wear, protective equipment and equipment variants.
  4. Train operators to stop and escalate when the physical state differs from the guide.
  5. Compare quality, time, downtime and safety with the prior method.
  6. Expand only after the content-maintenance workload and unit economics are visible.

The hard asset is trustworthy context

The round gives Squint resources to develop software and support more sites, but the durable value will depend on the quality of the industrial context captured around each asset. Models and phone interfaces can be copied more easily than a validated library linked to real equipment, permissions and outcomes.

For manufacturers, the relevant purchasing question is therefore not whether AI can display an instruction. It is whether the combined system can keep thousands of instructions current, retrieve the right one under plant conditions, incorporate worker corrections and demonstrate operational benefit without weakening safety authority.