Looq

Where Looq sits

Looq is the odd one out in most AI tool catalogs, and that is precisely why it is interesting: it is a hardware-enabled survey platform, not a pure software product. The San Diego company — founded in 2021 by CEO Dr. Dominique Meyer and chief science officer Shreyas Niradi, emerging from stealth at Distributech in February 2024 with $2.6M in seed funding ($3.75M total including the pre-seed) — sells a handheld photogrammetry camera (qCam) plus a cloud processing pipeline (qAI) and a web viewer (qApp) that together produce survey-grade 3D point clouds and 2D geospatial datasets from ground-level imagery. No LiDAR scanner, no drone, no aircraft. The target customers are land surveyors, civil engineers, transmission and distribution engineers and utility asset managers. What I look for in this category is whether a capture tool can deliver accuracy numbers a licensed surveyor will accept, and whether the workflow survives contact with real field conditions.

What it actually does

The workflow has six stages. Setup: initialize GPS, acquire satellite lock, place ground control points. Capture: walk or drive a route with the qCam recording everything from one screen, tracking your path, tagging features, with automatic people and vehicle filtering. Process: upload to qAI, where PPK post-processing and photogrammetry produce georeferenced 3D models and 2D composites, and AI segmentation classifies ground, buildings, vegetation, poles, wires, signs and more while extracting semantic and geometric features. Align: tie the model to real-world control — visual targets or PK nails — with residuals shown live and a PDF accuracy report exported. Draft: refine geometry, draft points and lines in 3D, assign field codes. Export: deliver imagery, point clouds and vector linework in GeoTIFF, LAZ, SHP, DXF and CSV into CAD and GIS workflows.

The qCam hardware is serious kit: a 20 MP multi-camera array with a 270x150-degree field of view rated up to 100 MP/sec, survey-grade GNSS with IMU, an onboard AI processor, IP67 dust and water resistance, 2 TB of encrypted onboard storage per capture, and a 99 Wh TSA-compliant V-mount battery lasting 4-5 hours in the field. It is handheld by default but vehicle-mountable. A specialist module, qPole, converts photographs into engineering-ready 3D pole models for utility work — the kind of deliverable used for pole-load analysis and phase tracing.

The accuracy claims are the product's spine, and they are specific. Out of the box with CORS and other corrections, Looq publishes global accuracy of about 3 cm horizontal and 6 cm vertical, with relative accuracy under 5 mm. Local base station PPK tightens global accuracy to roughly 2 cm; control point alignment gets under 1 cm. The company's internal tests against terrestrial LiDAR and licensed-surveyor markers report relative accuracy of 5-10 mm and global 3-5 cm RMS with a reference station within 10 km under ideal GNSS conditions — a vendor-run test, clearly labelled as such in their documentation, and exactly the kind of number a buyer should verify on their own control before trusting it on a certifiable as-built.

My testing notes

I approached this as a capture-and-deliver workflow review rather than a lab accuracy test, since the qCam is leased-with-hardware rather than downloadable. First, the capture ergonomics are well above what I expected from a first-generation device. The single-screen capture view with live path tracking and automatic people-and-vehicle filtering addresses the two things that ruin photogrammetry walks: losing your route pattern and capturing bystanders' faces into a dataset. Looq's own field run-rates — roughly 250 poles per day on foot and about 450 per day mixing vehicle and walking — match the pace of a crew that is capturing rather than fussing with equipment.

Second, the control-point alignment step is where the 'survey-grade' claim either holds or collapses, and Looq's design of this step is the most credible part of the product. You can preview expected accuracy before adjusting, mark points as control versus check, and watch residuals change live when you modify points, then export a PDF report. That is the workflow a licensed surveyor expects from adjustment software, transplanted into a photogrammetry pipeline. In the numbers Looq publishes, the tiered structure — 3 cm out of the box, 2 cm with a local base, sub-centimetre with control — is honest about the cost of each accuracy tier rather than quoting the best case as the default.

Third, the semantic layer is what differentiates it from a bare photogrammetry stack. The classification of poles, wires, signs and vegetation turns a pretty point cloud into an asset inventory, and qPole's engineering-ready pole models target analysis workflows directly. Looq's claim that qPole could save US grid engineers 19 million hours a year is a vendor projection — I would not repeat it as measurement — but the underlying workflow of one survey serving asset database updates, pole-load analysis and as-built records is a genuine efficiency argument.

What it does not do

This is not software you try tonight. The free tier exists, but real capture requires qCam hardware, sold on custom quotes, and the one third-party price page I found listing Pro software at $29.99/month is the only published number in an otherwise quote-driven sales process; another listing showed a headline figure above $30,000 that appeared to be a hardware bundle. Budget conversations happen with sales.

Accuracy is conditional, and the documentation says so: 3-5 cm RMS assumes ideal GNSS conditions and a reference station within 10 km; urban canyons, heavy canopy and long baselines degrade it. There is no LiDAR, so anything the cameras cannot see — the underside of a crossarm, the back side of a pole across a ditch — needs supplementary capture; Looq's own FAQ admits walking crews sometimes need to cross the road to get pole tags. Capture speed is also hardware-bound today: above roughly walking-pace collection the company says faster capture awaits future software updates. The company is young — 15 employees as of the 2024 launch coverage, $3.75M raised — which is a durability consideration for a product whose hardware you build a field program around. Data is stored and processed in the United States, which suits US utilities and will need review for EU teams.

I also checked the post-processing turnaround question, since cloud processing is a common bottleneck in capture platforms. Looq's documentation describes upload-to-cloud processing as part of the standard flow and the survey workflow shows it as an automatic stage rather than a manual job, but the company publishes no fixed processing-time SLA in the materials I could reach — for a large corridor project I would confirm turnaround expectations during the pilot rather than assume same-day results. The US-only data processing location is worth raising early for international infrastructure owners too.

Verdict

Looq is the most credible no-LiDAR survey-capture workflow I have reviewed, and the accuracy tiering is refreshingly honest for the category. I rate it 3.7 on the rubric: core functionality is deep for its niche — capture, PPK, semantic extraction, CAD-ready export in one chain (the 40% pillar), value is hard to judge without hardware quotes but the software-side pricing is low for what the pipeline delivers (20%), usability in the field design is good though the office tooling assumes surveying literacy (20%), the ecosystem is young — a small team and a narrow integration surface (10%), and the vendor-run accuracy tests plus early customer deployments are positive but early signals (10%). For utility and civil teams that need fast, safe ground capture between full survey campaigns, Looq deserves a pilot on your own control points. For anyone needing certifiable boundary surveys, treat it as a complement to licensed survey, not a replacement.

Sources

• https://looq.ai/about/
• https://looq.ai/hardware
• https://looq.ai/survey-civil
• https://support.looq.ai/welcome-to-looq/frequently-asked-questions
• https://support.looq.ai/welcome-to-looq/what-is-the-looq-platformuntitled-article
• https://docs.looq.ai/Looq%20AI%20-%20Frequently%20Asked%20Questions
• https://www.businesswire.com/news/home/20240227335776/en/Looq-AI-Launches-AI-Enabled-Digital-Twin-Platform-at-Distributech-2024-Secures-%242.6M-from-Top-VCs
• https://www.sdbj.com/technology/taking-an-ai-look-at-infrastructure/
• https://www.finsmes.com/2024/02/looq-ai-raises-2-6m-in-seed-funding.html
• https://www.venturebanc.com/news/closed-2m-seed-for-san-diego-survey-technology-platform
• https://rightaichoice.com/tools/looq-ai
• https://alternativeai.tools/tools/looq-ai