UpSmith, a Dallas company founded in 2022 that sells AI software to home-services contractors, has raised a $10 million Series A led by SemperVirens, Dallas Innovates reported on 6 October. Pulse 2.0 reported the round the following day with the same investor list. The company’s own announcement page confirms the amount — “$10 million in new capital” — but carries no date and names no investors.
Both outlets list the participants as Hannah Grey Ventures, Asymmetric Capital Partners, CIV, Synetro Group, the WaterStone Impact Fund anchored by Tim Tebow, Colin M. Evans of Sandwith Ventures, Deason Capital Services, and OpenGov co-founder Zachary Bookman. Allison Baum Gates, a general partner at SemperVirens, and Rob Biederman, managing partner at Asymmetric Capital Partners, join the board.
Dallas Innovates describes it as the company’s largest round to date. Dallas Innovates puts the earlier financing at a $5 million seed in 2024 and a $3.3 million pre-seed, which its 2024 report dates to October 2022 — placing the disclosed total at $18.3 million across three rounds.
The pivot is the story
UpSmith in January 2024 and UpSmith in October 2026 are selling to the same customer against a different problem, and the shift is documented well enough to be worth reading closely.
When the company raised its $5 million seed, Dallas Innovates covered it as a workforce story. The product was Boost, a productivity app that gamified technician performance: it plugged into a contractor’s existing systems and awarded “UpSmith points” for behaviours that created value — the example given was presenting a homeowner with four service options — redeemable in a rewards marketplace for tools, prizes or extra holiday. The framing around it was the skilled-labour crisis. The company cited more than 1.3 million open positions in construction and manufacturing, and described sourcing, screening and training candidates, including veterans and recent school leavers, then placing them with employers, claiming it could train a technician in under 90 days.
The 2026 product points at the other side of the same business. UpSmith now connects to a contractor’s ServiceTitan database — Pulse 2.0 describes the company as a certified ServiceTitan ecosystem partner — and runs AI agents over the operational history already sitting in it: stale estimates, missed calls, overdue maintenance, expiring memberships, dormant customers. The agents then act, confirming appointments by text, responding to leads, following up on estimates and running membership and remarketing programmes. A voice agent for inbound calls and a unified customer timeline are the next additions. The company reports more than 200 home-services operators as customers and, per Dallas Innovates, employs about fifteen people, mostly in Dallas, roughly half in engineering and product.
Analysis. The move from labour supply to revenue recovery is a rational one, and not only because selling software is easier than placing people. A contractor’s willingness to pay tracks the directness of the link to cash. Boost asked an owner to believe that a points marketplace would change technician behaviour, which would in turn change revenue — two inferential steps. The current product asks an owner to believe that estimates already quoted and not closed represent money on the table. One step, and the baseline sits in a database the owner already pays for.
It also explains the ServiceTitan dependency, which cuts both ways. Building on top of the system of record puts a contractor’s data within immediate reach — and puts the company within immediate reach of being replicated by it.
The customer figures, as published
These are company-reported and vendor-published, and each is attached to a named operator, which is better than the sector norm:
- HomeTown Services, a Texas contractor operating across five states, is credited with more than 5,000 completed jobs recovered from unsold estimates this year. Ray Pickens, its chief operating officer, is quoted saying, “The database was always the asset” — his point being that staff could not work it fast enough unaided.
- Five Star Home Services in Ohio, which UpSmith’s site describes as a $50m-plus operator, reports a 34% reduction in cancellations.
- Medley, a Carrollton, Texas HVAC and plumbing contractor, reports $147,000 in new revenue over three months and cancellations down 25%.
Treat these as the vendor’s strongest cases rather than a mean. Cancellation rates and estimate-close rates are both heavily dependent on how a given shop ran its follow-up before, and a contractor with a disciplined callback process has less to recover than one without.
Wyatt Smith, UpSmith’s founder and chief executive, says the plan over the next 12 to 18 months is to “ramp our investment in building world-class AI products for our customers”, and frames the thesis as a conviction that “AI should power the people who power America.” Baum Gates, for SemperVirens — a firm that describes itself as backing AI-native companies in the economy’s largest and most complex service markets — makes the familiar argument — “The skilled trades are the backbone of the American economy” — and says the technology built for these businesses has historically lagged.
Where this sits
Analysis. Home services is adjacent to construction rather than inside it: UpSmith’s buyers are HVAC, plumbing and electrical shops doing service and replacement work in existing buildings, not contractors putting up new ones. It is covered here for the same reason Probook’s $40 million from Sequoia and Andreessen Horowitz was in June — the trades have drawn a steady run of construction-adjacent AI rounds this year, and the specialty contractors in question overlap with the mechanical and electrical firms that also work commercial projects.
The two rounds also mark out different bets on where the trades’ software consolidates. Probook raised four times as much to build an operating system around the dispatch board — the scheduling core of the business. UpSmith has taken a smaller cheque to sit beside the system of record and monetise the customer history, which is a thinner wedge but a cheaper one to insert, since it requires no rip-and-replace.
Worth holding against both is ServiceTitan’s September report on a survey of 1,020 commercial contractors, in which 62% had tried AI and, among those using it, 15% could demonstrate clear ROI. UpSmith’s pitch lands directly in that gap: recovered revenue is one of the few AI outputs in the trades that a contractor can count without a methodology argument. Whether it holds up across 200 operators rather than three named ones is the thing $10 million is being spent to find out.
