The deck, in-page.
Skim the round here. Open full-screen, download a copy, or read the supporting documents below the viewer.
Click slide or press → · ← · Space
Supporting documents
Atlas United is raising a pre-seed round to take a working, USPTO-Filed peril intelligence platform from solo-founder traction to the first enterprise pilots and the first engineering hires. 282.9 million U.S. addresses indexed across 54 perils — every cell populated, no nulls — calibrated against 22 years of validated severe-event data, served from a single audit-ready API.
Skim the round here. Open full-screen, download a copy, or read the supporting documents below the viewer.
Supporting documents
For thirty years, peril risk in insurance has been priced by a handful of catastrophe-modeling vendors selling proprietary black-box scores. Carriers paid seven figures. Regulators got vendor whitepapers and trade-secret disclaimers. Underwriters got a number and were told to trust it.
Atlas United is built on the opposite premise: a deterministic, address-level peril model can be transparent, federally-sourced, cryptographically verifiable — and still be the most accurate way to price risk. One model. Fifty-three perils. 282.9 million addresses. Zero null cells across the matrix. A single API call returns the grade, the score, the reason code (patent claim 30), the calibration vintage, the model version, and a signed Merkle receipt of every source record used.
The market is moving. State regulators are pushing for explainable rating factors. Reinsurance treaty pricing is increasingly tied to model auditability. MGAs and parametric programs need per-address peril intelligence that survives litigation discovery. The incumbents cannot offer this — their moat is the opacity. Ours is the federal source admission and the cryptographic receipt.
Every peril layer is calibrated against public-federal sources only — NOAA storm events, FEMA NFHL flood zones, USGS faults, NIFC fire perimeters, USDA soil, FEMA NRI. No proprietary commercial data lock-in. Carriers and regulators can independently re-derive every score from the same public records.
The Address-coordinate-indexed Exposure & Peril Matrix (AEPM) was filed with the U.S. Patent and Trademark Office on May 20, 2026 as a §111(b) provisional application — establishing priority over the dual-radius computation, dual-horizon forecast retrieval, deterministic non-null peril matrix, binding-state finite-state machine, and Merkle-signed verification receipt.
Every API response carries an Ed25519 signature, a content hash, and a Merkle root over the federal source records used. Audit-ready by default. A carrier under regulatory scrutiny can prove which NOAA event, which FEMA zone vintage, and which forecast model-run produced any given grade — months or years later.
282.9 million addresses are indexed today, 54 perils per address, zero null cells. The peril matrix is partitioned and pre-warmed for sub-100 ms cold latency on the fast endpoint. The forward forecast layer refreshes hourly off NOAA HRRR and GFS. A new competitor entering this category starts with months of pipeline engineering before their first calibrated address.
Provisional patent filed May 20, 2026 covering the system, method, computer-readable medium, dual-radius exposure computation, dual-horizon forecast retrieval, deterministic non-null peril matrix generation, binding-state finite-state machine, and Merkle-signed verification receipts. 30 claims. Conversion to non-provisional scheduled within the 12-month statutory window.
Atlas United is pre-revenue on the enterprise side and operating a live production stack. The full peril modeling pipeline, the partitioned address cache, the verification-receipt scheme, and the live admin demos were built and shipped over the past several months.
Isolated insurance-api service on port 8082 returning per-address grades, scores, and reason codes across 54 perils. Decoupled from the forecast stack so carrier traffic is unaffected by long-horizon model runs.
NOAA HRRR short-horizon and NOAA GFS long-horizon forecast layers indexed at the address level. Forward-leads cache rebuilds hourly with auto-purge of past storm windows.
Every API response carries an Ed25519 signature over a Merkle root computed from canonicalized federal source records, scoring configuration, and model-run identifiers. SHA-256 leaf and node hashing per the filed patent specification.
Speed Demo (live cold-call latency feed), Full Carrier View (46-peril expanded grade view per address), and the StormCast lead-forecasting product running against the same underlying federal data.
Atlas United is raising a pre-seed round to convert the working product into the first paid carrier pilots and to make the first engineering hires beyond the founder. We are running lean on Microsoft for Startups Azure credits and a single-operator stack today. The next twelve months are about commercial validation, not rebuilding what already works.
First three hires — a database / backend engineer to own the partitioned peril cache, a GIS / pipeline engineer to expand the federal source catalog, and an SRE for production observability and on-call.
Insurance subject-matter hire (former carrier or MGA underwriting), pilot enablement, regulatory and licensing review, and the early carrier integration work.
Scale-out of the production database tier, snapshot and disaster-recovery posture, and 18 months of operating runway through the non-provisional patent conversion and first pilot revenue.
Atlas United is solo-founded by Valentin Yslas. Built and shipped the full stack — peril modeling pipeline, 282.9 million-row partitioned address cache with zero-null guarantee across 54 perils, dual-horizon forecast retrieval, Merkle receipt scheme, carrier-facing API, and the live admin demos — over the past several months as a single operator. Authored and filed the provisional patent (USPTO Application #64/070,786) directly with no outside counsel.
Operates a multi-server production environment across Azure (database + API), Hostinger (frontend + admin), and supporting infrastructure for the consumer brand. Background in operating real-world businesses end-to-end — building the product, the data, the API, the frontend, the deployment posture, and the customer conversations.
We respond to inbound within one business day. A pre-seed conversation can be scoped on a single call, and we can run a live API demo over screen-share before, during, or after.