The HVAC technician has already diagnosed the problem—a faulty compressor, maybe two hours of work—but now comes the part of the job that nobody wants to talk about. Photographing serial numbers. Filling out warranty forms. Logging equipment details into three different systems. Texting the customer. According to 2025 research from Salesforce, field service technicians burn roughly 30% of their hours on documentation and administrative tasks. That's more time filing paperwork than actually turning wrenches.
It's a grinding inefficiency that San Francisco-based Kebra thinks it can solve with AI agents that handle the administrative slog—filing warranty claims, syncing notes across platforms, even nudging forgetful techs via text when they leave a job site without snapping a photo of that all-important nameplate.
The startup, which went through Y Combinator a couple years back, is making a straightforward pitch to an industry not exactly known for its love of new software: let us automate the stuff you weren't documenting properly in the first place.
The System: Prompts, Photos, and a Digital Memory
Kebra's core product revolves around what the company calls "Field Capture." It's designed to prompt technicians while they're still on site—asking for equipment photos, serial numbers, diagnostic readings, anything that might otherwise get forgotten in the rush to the next appointment. Miss a detail? The system fires off an SMS reminder before the tech drives away.
Once a job wraps, the platform's AI agents kick in. The company's marketing materials show examples of these digital workers filing a Trane warranty claim (one case illustrated a $412 recovery), pushing job notes into ServiceTitan, generating QuickBooks invoices, reordering parts from suppliers like Johnstone Supply, and updating customers via text. Whether those integrations represent formal partnerships or API connections cobbled together by Kebra isn't entirely clear—partner certifications aren't listed publicly.
There's also a "Company Brain" feature, which is essentially a queryable database designed to capture institutional knowledge. The veteran tech who can diagnose a quirky boiler model by sound alone? Kebra wants that expertise accessible to the whole team, not retired along with him.
Who's Behind It
CEO Luca Hadife previously worked on multimodal AI systems at Microsoft and spent time at Amata, an early-stage startup. He studied data science and machine learning at UC Berkeley, where he also dabbled in student government. CTO Mohamed Hosny taught computer architecture and signal processing in Berkeley's EECS department before moving to Arista Networks, where he worked on data center switches.
It's a two-person operation, at least as of August 2026. The founders announced their launch on Y Combinator's platform, noting they were working with field service companies in Florida, Pennsylvania, and California. Customer names? Case studies with hard numbers? Those haven't surfaced yet.
A Crowded, Competitive Space
Kebra is hardly alone in trying to ease the documentation burden. The field service software market has become something of a feeding frenzy for AI-driven automation.
XOi, now majority-owned by private equity giant KKR after raising a substantial growth round recently, offers visual documentation tools and AI-generated work summaries. PTC rolled out ServiceMax AI with similar goals—automating scheduling and documentation. Even ServiceTitan, one of the platforms Kebra integrates with, has been layering in AI features like Field Assistant and virtual agents.
Microsoft has staked a claim here too, alongside TrueContext (formerly ProntoForms) and a handful of vertical-focused startups. Industry reports have flagged AI-supported workflows and structured documentation as a dominant trend, which means Kebra is operating in a space where the big players are already circling.
What It Costs, Who Can Use It
Kebra's website invites interested companies to book a 30-minute demo. Pricing? Not listed. That's fairly standard for enterprise software still finding its footing, though it does make comparisons difficult.
One detail worth noting: the company's privacy policy states that Kebra—operated under the legal entity Third Eye AI, Inc.—doesn't use customer emails, files, or other content to train generalized AI models. For service companies wary of handing over proprietary diagnostics or customer data, that's probably a selling point worth highlighting in sales calls.
The broader question is whether a two-person startup can carve out meaningful space in a market where incumbents like ServiceTitan and deep-pocketed entrants like Microsoft are already embedding AI into their platforms. Kebra's bet seems to be that smaller, focused tools—ones that don't require ripping out existing systems—can still win over contractors tired of bloated, hard-to-navigate enterprise software.
For now, the pitch remains simple: give techs their time back. Whether that resonates enough to build a sustainable business in a crowded market is the question every Y Combinator company eventually has to answer.
