See the whole picture.
Prove every line of it.
VALIS turns scattered open-source signals into an auditable intelligence picture — collected, correlated, reasoned over, and packaged for a decision. One workspace, many analysts, working the same problem in real time, online or cut off in the field — and now across many federated headquarters, encrypted in depth.
Vast Active Living Intelligence System
Building the picture
Everything begins with an area of interest and the layers you pull into it. VALIS keeps collection and analysis separate from the feeds themselves, so adding data never means changing code.
Type a place, a lat/lon pair, or an MGRS reference into the search box; VALIS resolves it to a bounding box and flies you there. This AOI is the lens for everything that follows — collection, analysis, offline bundles and watches all key off it.
The Data Layers panel lists every feed your team is granted — aircraft (live ADS-B), vessels (AIS), earthquakes, weather, transit, infrastructure and more. Toggle a layer and press Collect; entities land on the map, graded for reliability, and shared with your whole team instantly.
Making sense of it
A map full of dots is data, not intelligence. VALIS’s analysis tools turn what is there into what it means — and every one is deterministic: given the same data it gives the same answer, and it can always show its working. That is what makes the output defensible.
Meaningful pairings — close in space, close in time, co-incident — plus density clusters and same-entity candidates.
Track history, reachability, viewsheds and routing turn a position into a projection.
Multi-source convergence, severity-ranked and weighted for recency and source reliability.
Reading documents — who, when & what
Much of intelligence arrives as text. VALIS’s Dossier reads a document’s content and pulls out the entities that matter — coordinates and grids, people, organisations, dates, money, comms identifiers and case references, each with a confidence score. Paste text or drop a .txt, .docx or .pdf; every extraction is highlighted in the source, so you verify each finding in context before trusting it.
No model to host and no service to call, so it runs identically inside an air-gapped enclave — exactly where document intelligence is often needed most.
Each extraction traces to the rule that found it and the text it came from — not the opaque, unrepeatable output of a model you cannot inspect or defend.
A rule engine can only find what is genuinely in the text. It cannot fabricate a name, a grid or a date that was never there — the failure mode that makes LLMs unsafe for evidence.
It is also deliberately precision-biased: it would rather miss a weak candidate than clutter your graph with false ones. The result is document intelligence you can put in front of a decision-maker and defend line by line.
Staying honest and unbiased
The hardest part of the job isn’t finding data — it’s not fooling yourself. VALIS builds the discipline of structured analysis into the tool, so your assessment is the product of a method, not a hunch.
Lay out every plausible explanation side by side, score each piece of evidence against all of them, and let the matrix show which hypothesis the evidence fails to disprove — the opposite of cherry-picking.
A timestamped journal on every investigation. Reference entities inline and they auto-link. Chain-of-reasoning beside chain-of-custody.
Standing watch while you work
You can’t stare at every feed forever. Watchlists and tripwires turn VALIS from something you pull answers from into something that tells you when the situation changes.
Alert when a specific hull, tail number, callsign or named entity reappears — even after going dark.
Alert when anything — optionally from named sources — enters an area you care about.
Alert when fusion raises a new indicator at or above a severity you set.
Hits arrive live at the alert bell — no polling, no refreshing — shared with your whole team. Acknowledge them, jump to the entity, or fire them onward via a secure webhook.
The target package
Analysis that stays on your screen helps no one. VALIS assembles your work into a structured intelligence product — an INTSUM model and a briefing deck — that a decision-maker can act on, with the evidence graded and traceable beneath every line.
Working as a team, at the speed of the problem
The unit of collaboration is the workspace — a shared compartment where a team sees the same picture, in real time, with strict isolation from other teams.
When any analyst collects, tags or links, it appears on every teammate’s map within moments. No “send me your file.”
Enforced in the database itself. You only see the workspaces you belong to; another team’s data doesn’t exist from where you sit.
Deconfliction: open any entity and see who on your team has touched it and when — plus a privacy-preserving hint if another team is working the same object, so two desks don’t unknowingly duplicate a target.
Offline and in the field
The environments where intelligence matters most are exactly where connectivity fails. VALIS is designed to keep working when the link drops — and to reconcile cleanly when it returns.
Take it offline before you lose signal: your AOI, movement history, your team’s tags, links and investigations, plus the basemap, cache into the browser. Then work as normal — the analysis engines run right in the browser. Sync when you can: one action pushes your work up and pulls teammates’ changes down, idempotently, with clean conflict review.
Federation — many headquarters, one picture
Real intelligence organisations are not one site — they are fleets, coalitions, forward teams and headquarters, connected by links that are often slow, intermittent, or absent. Federation lets many independent VALIS headquarters work a problem together as if they were one system, while each remains a complete, self-sufficient VALIS in its own right.
Each headquarters runs the full VALIS and is authoritative for its own work. It never waits on anyone else to let its analysts work.
Sites exchange only what has changed, so the picture converges across even a poor link — and both sides can prove they now match.
There is no central server whose loss stops the mission. Lose any site — even the primary — and the rest continue undisturbed.
On screen, a small federation pill names your headquarters, shows how many peers are reachable, and displays the digest — a fingerprint of your whole workspace. When two sites show the same digest, their pictures are not merely “in sync” — they are provably identical.
Protecting your intelligence
Intelligence that moves between sites and sits on machines in the field needs protection at every layer — not one lock, but several, so a single failure never hands an adversary the whole picture. VALIS encrypts in depth.
The datastore, its backups and its logs sit on encrypted volumes. A seized laptop or lifted disk yields ciphertext, not intelligence.
Links between headquarters are mutually authenticated and encrypted, and every batch of changes is signed by its origin — so a site can prove a change is genuine and untampered.
The most sensitive fields are encrypted before they ever reach the database, so a compromised database account sees only ciphertext. Keys never sit beside the data they protect.
Agents, export & external systems
VALIS isn’t a walled garden. Through the Model Context Protocol, an AI agent can drive VALIS with the same toolset an analyst uses — and, crucially, an agent is just another analyst: it authenticates and is authorised exactly like a human, confined to its workspaces, gated by the same grants, subject to the same rate limits, and fully audited. No privileged back door.
When the decision lives elsewhere, your workspace picture exports cleanly to STIX 2.1, MISP and KML — carrying only what you’re cleared to see, respecting handling caveats as it leaves, and recorded in the audit trail. A documented REST API serves non-AI integrations over the same authenticated, compartmented channel.
A worked mission
How the pieces fit on one problem — assess unusual maritime activity in a strait over the last week — from empty map to delivered package, across a federated task group.
Search the strait, create a workspace, bring in teammates. Toggle vessels, aircraft and org feeds; collect the last seven days, Admiralty-graded and shared live.
Drop a geofence and a HIGH indicator watch. Run correlation; a proximity-and-time link surfaces a candidate rendezvous. Tag both vessels.
Movement shows a loiter; fusion raises a HIGH indicator; Cloud Atlas confirms a hull. Build an ACH matrix and log the reasoning.
Generate the INTSUM and deck, apply the caveat, hand it up. Leave the watchlists running for the next shift.
Notice what carried through: a shared picture the team built together, deterministic analysis you can defend, reasoning captured as you went, and a product whose every claim is backed by graded, traceable evidence — produced in an afternoon, resilient to a dropped link, federated across sites, encrypted in depth, and open to automation under supervision.
Quick reference
Place / coordinates / MGRS sets your area of interest. Assemble the INTSUM and briefing deck. Live watchlist hits for the workspace. ONLINE / OFFLINE / ENCLAVE with Take offline, Sync, and — across sites — federation status. Switch compartments, with your role shown beside each. Users, membership, feed access, the no-code feed builder, and peer enrolment & keys for federation.
Collect → Correlate → Fuse → Test → Package — feeds and docs to the map, find links, raise indications and warnings, ACH and notebook, then INTSUM and deck.