enigma_data

Agent reliability is an infrastructure problem, not a model problem. That’s actually a good thing. It just means you have to start with reliable data. A well-reasoned KYB decision drawn from stale filings is still wrong. A confident match against an unresolved entity is still a false positive. before tuning prompts, audit your data sources. Before adding agentic logic, evaluate resolution quality. In other words: fix the stack from the bottom up, not the prompt from the top down. And learn the five elements of the KYB stack on the Enigma Knowledge hub: enigma.com/resources/know…

When top-tier models use web search to run KYB checks, the result is a coin flip, often worse. That's the agentic baseline, even when you throw lots of tokens at the problem. When the Enigma KYB Agent for Google Cloud runs the same checks against our verified business identity graph, it gets 20% higher accuracy at one-tenth the inference cost. The math holds because the agent isn't burning tokens working out which business it's looking at. Identity is resolved before the agent even starts: brand to legal entity, every variation, every jurisdiction, every filing, plus operating signals to know whether the business is real. Built on Vertex AI Agent Engine, Gemini Flash, and Cloud Run. Submit a single business through chat, or run a batch through a spreadsheet. Every check returns its reasoning, grounding every decision in our verified identity graph. Learn more: enigma.com/resources/blog…

Shell companies are designed to look smooth to standard KYB checks: the legal entity exists, the address resolves, the state filings are current. Green light. That's where shell networks slip through, because basic verification confirms that the business is on record. The harder question is whether the business does anything in the real world, or connects to anything else that does. But you can find the data signature of shell companies in other ways: network signals, operating gaps, and the connections no individual entity check sees. Wanna see the entity-level signals that distinguish a real business from a hollow one? enigma.com/resources/blog…

Over 100,000 businesses in the U.S. have hit their tenth birthday on paper without ever opening a location, generating revenue, or showing up anywhere on the open web. We call these businesses “zombies” for good reason: they cluster at the same handful of addresses, register in coordinated waves during specific years, and consistently reanimate the same few corporate structures. A registration on file is a cheap proof of existence. Confirming that a business actually operates is harder: its operating signals cannot be stamped and filed into existence, and much of the KYB field has built its workflows to consult the filing rather than the signals. We mapped the gap between filing and operation, and estimated what share of each state’s “active” businesses are really just zombies. → enigma.com/resources/blog…

Did you know that people can fill out business registrations with pretty much any information they want? State business registries don't ask for ID, signatures, or any verification of the people named in a filing. Pay the registration fee and they accept whatever you type in the form. Of course, any system that can be abused, will be abused. One LLC lists Mickey Mouse and Minnie Mouse as officers. The state of Georgia treats this corporation as a real, active business. That’s why Enigma's knowledge graph uses operating signals to verify that a company is actually transacting business. Otherwise, Mickey Mouse comes back valid. We searched 103 million registrations across all 50 states to find the most egregious cases: → enigma.com/resources/blog…

“Identity verification” is a single phrase covering two fundamentally different problems. To verify a person, you need a government-issued ID, a selfie, and a liveness check. The inputs are a document and a face to tell you: is this the individual the document describes, and are they physically present right now? Verifying a business is a different scale of problem. The data required spans government registries across dozens of jurisdictions, corporate filing histories, ownership structures that shift over time, and sanctions watchlists that can change daily. As agents take on more of the transaction layer, they encounter both problems at once. Approving an application means knowing the individual submitting it. Onboarding a vendor means knowing the business behind it. That requires a different data layer underneath: a verified business identity graph.

Enigma's card-backed revenue signals are now inside Clay, giving GTM teams real transaction data from actual payment networks. Major financial institutions already rely on this data for compliance. Now GTM teams can use it for growth. prnewswire.com/news-releases/…

New esoteric dataset just dropped

Enigma Gov Archive is live. Billions of government records across thousands of datasets, growing continuously. OSHA violations, EPA enforcements, health inspections, business licenses, UCC filings, building permits, campaign finance disclosures. Federal, state, and local agencies. All in one database. Agent-ready from day one: MCP, API, and direct integrations. The business identity layer your agent needs to reason on government records instead of guessing from web results. Every data point traces to a specific agency, dataset, and filing. The web results most AI agents use aren't sourced. They aren't verified. They don't trace to a specific government filing. If your agent searches the web for business compliance data, it's guessing on the facts and reasoning on noise. With Gov Archive, your AI agent doesn't just search text. It searches a verified business entity, resolving every name variation, legal entity, and DBA through the same infrastructure that powers Enigma's KYB and compliance products. That's how one Gov Archive query returns records from many government databases, all matched to the same business. enigma.com/govarchive

Your KYB provider sees a registered agent address and sends it to manual review. Enigma sees a registered agent address and finds the physical business behind it. @tryramp put that to the test — 4,000 real applications, a 94% match rate, and a KYB integration that grew into a six-product fraud intelligence pipeline. Read the case study: enigma.com/resources/case…

We used Enigma's database of 31 million businesses to predict the March Madness tournament. The city with more wedding venues per divorce lawyer correctly called 71% of men's games last year. Explore hundreds of metrics and build your own bracket here: enigma-bracketology.vercel.app

Everyone knows that tax prep has the most extreme seasonality in American business. March revenues are about 18x June, so most of the 102,000 registered tax prep businesses open in January and close by May. But here's the thing — our data shows that October's average revenue per preparer is actually higher than any other month. 19M extension filers. Fewer shops open. Bigger returns. When you look closely at the seasonality of taxes, you see a story beyond springtime. Learn more about Enigma's data and what it can uncover at enigma.com

If you think Valentine’s Day is a goldmine for restaurants and jewelers, card transaction data tells a different story. Restaurants actually dip below their monthly average in February. Jewelers barely budge. Only florists get a real bump — and Mother’s Day is still bigger. Learn more about Enigma's Peak Revenue Month attribute at enigma.com/resources/blog…

State business records are official documents, right? Then why did Georgia approve Mickey Mouse as CFO and Minnie Mouse as Corporate Secretary of a registered business? Because registries don't verify. They just record. See the hundreds of businesses we’ve filed under fiction at enigma.com/resources/blog…