A garbage truck is not anybody’s idea of a glamorous technology platform. It growls, stops, lifts, and moves on. Yet it possesses a quality that makes software founders salivate: distribution. It travels nearly every street on a dependable schedule. Gavin Baum-Blake looked at that ordinary municipal machine and saw the beginning of a citywide sensing network.
City Detect, the Tuscaloosa company he co-founded and runs, mounts high-resolution cameras on public vehicles such as garbage trucks and street sweepers. Computer-vision models examine the passing view for visible conditions - graffiti, illegal dumping, overgrown lots, damaged roofs, abandoned structures. The result is a map that tells a municipal team where to look more closely.
The last phrase matters. Baum-Blake does not publicly describe the software as a judge. He describes it as a wider first look. A code officer verifies a detection before an intervention. A person brings context to the curb, including the local trash schedule, the difference between a mural and vandalism, and the delicate question of what to say when someone answers the door.
It is a practical proposition dressed in the unavoidable language of AI. City Detect’s real competitor, Baum-Blake has said, is the status quo: the windshield survey, parcel by parcel, clipboard by clipboard. A person might file roughly 50 reports in a week. The platform can produce thousands of insights in a day. The point is not to manufacture thousands of citations. It is to make limited public attention less accidental.
What a city buys, then, is not simply a camera or an algorithm. It buys a recurring picture of conditions that can be sorted by place, severity, district, and type. The old survey often captured a moment and began aging before the final spreadsheet arrived. A vehicle already following its route can return and show whether an issue remains, whether a notice worked, or whether limited repair money landed where officials intended. The glamorous word is prediction. The useful word is follow-up.
That distinction also suits Baum-Blake’s professional mix. An intelligence analyst learns that collecting a signal is different from understanding it. A lawyer learns that a fact acquires meaning inside rules and procedure. A founder learns, sometimes expensively, that a clever demonstration is different from a product somebody will budget for, train on, and use again next year. City Detect sits at the intersection of all three lessons. Its images matter only after they become an intelligible case for action.
The intended customer is stretched rather than dazzled. Code-enforcement departments deal with ordinary, persistent things: vegetation that grows back, furniture left at a curb, a roof that deteriorates slowly, a complaint line that reflects who knows how to use it. There is little theatrical about the work. Baum-Blake’s pitch works because it respects that plainness. The technology rides along, makes the first pass, and hands the public employee a shorter list with a clearer reason for each stop.
An unusual founder stack
Baum-Blake arrived at municipal computer vision by a route that feels assembled from three different résumés. He served in the U.S. Army as a signals intelligence analyst and Arabic linguist. He went to the University of Alabama School of Law, competed in mock trial, graduated in 2021, and became a licensed attorney. Before City Detect, he founded DontHaggle, a negotiation technology venture that gave him an early education in startup difficulty.
None of those experiences makes him the person training the vision models. City Detect’s technical origins lie with co-founder Erik Johnson, an economist and data scientist whose research examined how visible property condition connects to value. Baum-Blake’s contribution sits where technology meets institutions: operations, law, administration, financing, and the long persuasion required to get a public customer comfortable with a new tool.
The company’s early victories came in literal pitches. In 2022, City Detect won the $50,000 grand prize in the Edward K. Aldag Jr. Business Plan Competition. Later that year Baum-Blake won a $10,000 regional veteran pitch prize in Charlotte. The company launched commercially in 2023, raised a $2 million seed round in 2024, and announced a $13 million Series A in March 2026.
The numbers make a tidy staircase. The business underneath is less tidy. Municipal sales pass through budget cycles, procurement rules, legal review, privacy questions, public meetings, and staff members who have heard many promises from vendors. Baum-Blake’s advice to founders is appropriately unromantic: government partnerships take time; persistence, trust, and measurable impact matter.
The product is a boundary
One consequential City Detect design choice concerns what it refuses to care about. Its imagery blurs faces and license plates. Baum-Blake calls it a built-environment tool: identity is irrelevant to the problem being solved. The company has published a Responsible AI policy, joined the GovAI Coalition, and obtained an independent security compliance certification. Those commitments offer public customers something concrete to inspect before the first camera takes a route.
The boundary is not philosophical decoration. Consider a garbage can at the curb. On collection day, it is ordinary. Three days later, it may be part of an illegal-dumping problem. Spray paint can be a community mural or a tag. A model that merely recognizes objects has not solved the municipal question. It must reckon with schedules, sanctioned art, local rules, and uneven neighborhoods. Even then, Baum-Blake insists, the detection is the beginning of work rather than its conclusion.
That leaves code officers with the distinctly human part: verification, discretion, empathy, and conversation. It also gives them a chance to work proactively instead of responding only to complaints. Complaint volume is not a clean map of need. Some neighborhoods have active associations and residents who know exactly where to call. Others can carry equal problems in greater silence. A citywide survey can reveal both.
After the alert
Detection counts are seductive because they are large. Outcomes are harder and more interesting. In Stockton, California, City Detect’s platform identified more than 4,000 potential code violations across 2,500 properties in a week. The city used those findings in an education-first program. Eighty percent of residents who received a notice ultimately complied. The software helped locate and rank the visible issue; the municipal program determined what happened next.
Cleveland offers another kind of contrast. A comprehensive 2022 property survey cost $170,000, involved 40 officers, and took six months. City Detect says a single equipped vehicle can cover comparable ground in weeks. That comparison explains the appeal to a department with vacancies and expanding duties. It also explains why Baum-Blake keeps talking about capacity. The public employee remains responsible, but begins with a much more complete draft of the city.
The draft has already found uses beyond routine code enforcement. After Hurricane Helene, City Detect helped Greenville, South Carolina, perform a rapid storm-damage assessment. The 2026 financing is intended to support more engineering, municipal integrations, public-works products, infrastructure asset detection, and post-storm tools. Expansion will test whether the company’s careful boundaries survive new data, new departments, and new reasons to point a camera down a street.
A company that can travel without him
Founders like to discuss scale in dashboards. Baum-Blake recently found a more personal measure on a family trip to Ireland: 13 travelers, two babies, three rental cars, two breakdowns, and one new city each night. It was his first proper stretch out of office in a long time. From afar, he watched City Detect’s team continue without missing a step.
The anecdote is funny because founder freedom rarely arrives gracefully. More important, it mirrors the company’s pitch. City Detect exists so a municipal team can see more without asking every officer to drive every parcel. Baum-Blake now has to build an organization that can do more without routing every decision through its CEO. Both are exercises in designing attention.
His public persona remains that of an operator translating between worlds. There is the lawyer’s insistence on explicit policy, the intelligence analyst’s appetite for useful signals, and the repeat founder’s respect for the slow work of relationships. He credits accelerators for early introductions, investors for staying beside the company through difficult stretches, and public servants for applying the output with context.
The ambition is national, but the product stays stubbornly local. A roof is damaged on one house. A pile of tires sits beside one road. A mural belongs to one community. City Detect can make those small facts visible at city scale; it cannot decide what a city owes the people living beside them. Baum-Blake’s wager is that better sight creates room for better judgment. The camera does the looking. Public servants still have to see.