CCTV camera overlooking a construction site with AI overlays highlighting workers and equipment safety zones.

Ailytics Raises $7.4M to Turn Ordinary CCTV Into Industrial Safety AI

ByAIERA Blog Editorial Team

Key Takeaways

  • Ailytics $7.4M Series A vision AI funding: Singapore startup Ailytics has raised US$7.4 million in a Series A led by Tin Men Capital to expand across APAC, the US, and the Gulf.
  • Its product, Ailyssa, derives three-dimensional spatial understanding from a single ordinary 2D CCTV feed, so sites need no new cameras or sensor hardware to start.
  • A newer reasoning engine, AilyssaX, checks whether multi-step industrial procedures were actually followed, moving the system from detection toward procedure verification.
  • The company claims more than 400 deployments across 15 countries, up from 70 projects in four countries at its April 2024 pre-Series A. That figure is a company claim, not an independently audited number.
  • The money funds larger local teams in the US, Japan, Australia, and the Gulf, plus the company’s own domain-specific reasoning models, focused on harsh outdoor industrial environments.

A funding round for cameras that already exist

Most industrial AI pitches start with new hardware. Ailytics starts with the opposite argument: the cameras are already there. Construction sites, shipyards, warehouses, and airports have spent years mounting CCTV for security and record-keeping. Ailytics says its software turns those existing feeds into a real-time safety and operations layer, watching for danger as work happens rather than reconstructing it after an incident. The Series A, announced via PR Newswire on 25 October 2026 and carried by Channel NewsAsia on 26 October, was led by Singapore venture firm Tin Men Capital. Joining the round were SEEDS, part of SG Growth Capital, Safie, described as Japan’s largest cloud camera company, and one of Japan’s largest construction companies, which the release does not name. Family offices and existing investors also participated. The round follows a US$2.7 million pre-Series A in April 2024, also led by Tin Men Capital, putting the new raise at close to three times the size of the last one.

Two details are worth noting for what they do not say. The announcement gives no valuation and does not break down how much each investor contributed, and the Japanese construction participant stays unnamed, so its identity and any commercial commitment behind the investment remain unconfirmed.

How the Ailytics $7.4M Series A Vision AI Actually Works

The core technical claim is easy to state and hard to do well: take one flat camera image and recover enough depth and distance information to judge safety in three dimensions. A single 2D feed has no depth sensor, no stereo pair, and no lidar point cloud. Ailyssa must infer spatial relationships from the image itself, using scene geometry, the apparent size and position of people and machines, and learned patterns of movement through industrial spaces.

In practical terms, that spatial understanding is what turns a camera from a recording device into a measuring device. If the system can estimate where a worker stands relative to a suspended load, it can draw a danger zone under that load and flag anyone who enters it. If it can estimate a fixed radius around hazardous equipment, it can enforce an exclusion zone without new ground markings or worker tags. The company says the system works through rain, dust, and poor lighting, and with cameras that may be decades old.

This is also where the business model becomes clear. Ripping out cameras or blanketing a site with new sensors is expensive and disruptive, and it gives safety managers a reason to delay. Software that runs on the existing feed removes that barrier. The trade-off is accuracy: a single-camera depth estimate will never be as precise as a dedicated sensor, and performance will vary with camera placement, angle, occlusion, and weather. The release publishes no accuracy benchmarks or false-alarm rates, so buyers must take the performance claims on trust or run their own pilots. Vision AI applied to physical work is a broader trend, as shown by our look at a vision AI sorting machine, but the industrial safety version is stricter, because a missed detection can mean an injury rather than a mis-sorted item.

AilyssaX: from spotting hazards to verifying procedures

Simple detection answers one question: is something wrong right now. AilyssaX, the company’s newer reasoning engine, is aimed at a harder question: was the correct procedure followed, step by step. The company says the engine processes contextual scenes as they unfold and verifies multi-step procedures, producing more accurate and more meaningful alerts. On a real site, a lifting operation is a sequence: the area is cleared, the load is rigged, the exclusion zone is set, the lift is signaled, and the load moves along a planned path. A system that tracks the sequence can flag a skipped step rather than just a risky position, which matters because safety managers are drowning in alerts from systems that cry wolf. The caveat is that the release describes this capability in general terms only, with no specific procedures, accuracy figures, or third-party evaluation, so it should be read as a company claim awaiting customer evidence.

Where the system is already deployed

Ailytics began on Singapore public housing projects, and its early work grew out of a joint research project between the National University of Singapore and the Housing and Development Board. From that base, the company says it now runs large deployments at steelmaker NatSteel, shipyard operator PaxOcean, logistics company DHL, Rome’s Leonardo da Vinci International Airport, and the Changi Airport Terminal 5 build, where it says it watches more than 1,000 cameras. Those names span construction, maritime, logistics, and aviation infrastructure, all of which share large outdoor spaces, heavy moving equipment, and more workers than human supervisors can watch at once.

The deployment count deserves careful handling. The company reports growth from 70 projects in four countries at the 2024 pre-Series A to more than 400 deployments across 15 countries today, and it says its platform has flagged more than 5.7 million hazards. Both figures come from the company itself, no independent audit has been published, and the release does not define what counts as a deployment. Readers should treat 400-plus as the company’s own account of its scale, not as verified fact. The company also says customers report recordable incidents down 73 percent and safety non-compliances down 84 percent, but the release gives no sample size, baseline period, or methodology, so those numbers are directional at best.

What the money funds, and where it expands

Co-founder and chief technology officer Prateek Manocha said the round lets Ailytics build its own domain-specific reasoning models and keep the cost of running AI low as demand grows. Domain-specific models trained on industrial scenes should, in theory, beat general-purpose vision models on the exact situations that matter on a site, and low inference costs matter because the system watches video continuously.

Co-founder and chief executive officer Lenard Tan framed the geographic plan directly: the round strengthens local teams in the US, Japan, Australia, and the Gulf, with a focus on harsh outdoor industrial environments, a harder and less served market than indoor warehouse AI with its controlled lighting and predictable layouts.

Japan looks like the most structured part of the expansion. Ailytics and Safie signed a memorandum of understanding in April to explore pairing Ailytics analytics with Safie cloud recording, including reseller arrangements and pilot projects. Safie says it is used by 29 of Japan’s 30 largest construction companies, so if that partnership converts into distribution, Japan could become the company’s second home market. In the US, Tan said the target is mid-sized industrial firms that larger incumbents have priced out, a classic wedge strategy that avoids head-on competition for the biggest customers.

The safety numbers, and the limits to watch

The market context is sobering. The International Labour Organization estimated in November 2023 that work-related accidents and diseases kill nearly three million workers a year, though about 2.6 million of those deaths are from disease rather than accidents. Accidents account for roughly 330,000 deaths, and that smaller share is the one a site camera is aimed at. In Singapore, the Ministry of Manpower counted 36 workplace deaths in 2025, with construction accounting for 13, transportation and storage for seven, and manufacturing for four.

Several questions remain open. The release does not disclose pricing, contract lengths, or revenue, so the business model’s durability is unproven in public. Tin Men Capital’s Jeremy Tan cited robust unit economics and client churn below 1 percent, but that assessment comes from the lead investor, not from audited financials. The system’s false-alarm rate and its behavior when cameras are occluded or misaligned are unquantified in public, and the release does not detail where processing happens or how worker privacy is handled. Finally, expanding into four regions at once stretches a team that has just raised its first US$7.4 million round, so execution risk is real.

None of that diminishes the core idea. Turning existing CCTV into a safety layer is a genuinely attractive proposition, and the step from detection to procedure verification is the right direction for the field. The funding gives Ailytics room to prove the harder claims: that single-camera spatial understanding holds up outdoors, that AilyssaX reduces alert fatigue rather than adding to it, and that the deployment count converts into durable revenue. The next evidence to watch is not another funding announcement but customer results with named baselines, ideally from the Changi Terminal 5 build or the Rome airport deployment. This round also fits a wider pattern of focused, vertical AI companies raising mid-size rounds to serve one industry deeply, as seen in our coverage of Daytona’s $60M agent cloud raise. Ailytics has chosen industrial safety as its wedge, and the next two years will show whether ordinary CCTV can really carry that weight.

Sources

Similar Posts