StreetProof Joins NVIDIA Inception

The internet is good at repeating business claims. A company can publish a location, a service area, or a fleet size, and that information can spread through directories and AI answers long after it stops being accurate. What is harder to find is a record of what someone independently saw, when they saw it, and what that observation can reasonably support.

That is the problem StreetProof is working on.

StreetProof was founded by the operator of Calgary Garage Door Fix, a business that also appears in the index. We disclose that relationship because it is exactly the kind of fact a reader should know when judging an evidence system. A business owner understands the gap between doing work in the real world and trying to prove it online. But familiarity cannot be a shortcut: our own business is subject to the same evidence rules as everyone else. Paying for a membership, claiming a record, or knowing the founder cannot manufacture an independent observation.

StreetProof turns visible business presence—storefronts, signs, branded vehicles, and other public-facing signals—into reviewed, time-stamped physical citations. Each citation records a narrow fact about an observation, with provenance and limits attached. A wrapped truck seen in a city, for example, supports the claim that the brand had mobile presence there. It does not prove the company has an office there, operates throughout the city, or holds a particular licence.

That distinction matters. Search and AI systems are increasingly asked to answer questions about businesses, but a confident answer is only as good as the evidence behind it. We want them to be able to separate “a business says this” from “this was independently observed,” then see where the observation stops short.

Consider two questions a search product might ask: “Was this brand seen in this city?” and “Does this company operate from this address?” A branded van might help answer the first and tell you nothing about the second. A storefront with corroborating location evidence can support a different, narrower conclusion. If a system collapses both into one generic “business present” flag, it risks giving people the wrong answer with a convincing citation attached.

StreetProof makes those records available through a public website and machine-readable interfaces, including REST and MCP. Businesses can claim their identity and publish official information alongside the independent record. A claim does not create a physical citation or change how evidence is reviewed.

The index is already being read. In a public telemetry snapshot taken October 7, 4,140 of 4,219 external REST requests (98.1%) in the preceding seven days went to StreetProof entity or evidence endpoints. The chart below also shows successful external machine requests for seven completed UTC days. Those requests demonstrate retrieval of the index—not that a named AI provider cites it, recommends StreetProof, or relies on it in production.

Behind a short citation is a lot of hard engineering. StreetProof uses computer vision and other AI-assisted processing to find possible business signals in imagery, recognize text, match observations to the right entity, and screen sensitive details before evidence is retained or published. Human review and explicit inference limits remain part of the process. A model's guess is not a citation.

NVIDIA Inception gives member startups access to development resources, training, program offers, and opportunities to connect across NVIDIA's startup ecosystem. Those resources are relevant to a product that has to process visual evidence accurately and efficiently while keeping privacy rules intact. We intend to use the program to strengthen the technical work and learn from others solving difficult AI infrastructure problems.

Acceptance into Inception is membership in a startup program. It is not an NVIDIA investment, a product certification, or an endorsement of any StreetProof citation or claim. Those claims must stand on their own evidence.