Welcome to Krataxis
Krataxis is a geospatial intelligence fusion platform. In plain terms, it pulls in a continuous stream of open-source reporting (news wires, broadcast material, conflict-event data, humanitarian and environmental feeds, geolocated observations) and helps you turn that noisy, contradictory pile into something you can stand behind: corroborated, credibility-weighted intelligence with a clear trail that shows how you got there.
This knowledge base is your reference as you learn the workstation. It runs from your first sign-in through advanced workflows such as reporting, investigations, and outbreak tracking. You do not need to read it cover to cover. Start with the article that matches what you are trying to do, and follow the links from there.
The problem it solves
When you are tracking a global crisis, the hard part is not finding information. The hard part is that there is far too much of it, and none of it arrives with a label that tells you how much to trust it.
Think about what happens when a strike is reported. Three reports of the same strike might be one event witnessed by three independent observers. That is strong corroboration. Or they might be one press release copied three times, which is no corroboration at all, just an echo. A claim might come from an actor with a solid track record, or from one that habitually exaggerates. Sorting genuine signal from repetition and spin is slow, careful work. Doing it by hand for hundreds of items a day does not scale.
It is the same problem an auditor faces: a company's own statement that it paid an invoice is one thing; the matching bank transfer is another. You want both before you sign off. Krataxis does that receipt-matching for you, continuously, across everything that comes in. The time you spend is spent on judgement: deciding what is confirmed, what it means, and what to do about it.
The mental model
Almost everything in the platform follows the same five-stage arc. Hold this in your head and the rest of the interface makes sense:
| Stage | What happens | Who does it |
|---|---|---|
| Collect | Reporting is aggregated from trusted sources in near real-time and reconciled into one stream. | Automatic |
| Weight | Every item is scored for credibility across several independent factors. | Automatic |
| Assign | Items are resolved to the actors involved and routed to the battlespaces they're relevant to. | Automatic |
| Correlate | Related items are linked across space, time, actor, and topic. | Automatic, you confirm |
| Conclude | You escalate what matters into confirmed events and brief from the result. | You |
The first four stages run on their own, around the clock. The fifth — judgement — is yours, and the platform keeps a full record of every decision so your conclusions stay defensible. If a conclusion is ever challenged, you can walk back through exactly which items supported it and how they were weighted.
Notice the shape of that pipeline: it's a funnel. A wide mouth taking in everything, progressively narrowing as items are scored, routed, and linked, ending at the narrow spout where a human decides. Most of this documentation is really just explaining one part of that funnel or another.
The pieces you'll meet
A handful of terms recur throughout the platform. Don't memorise them now — every category explains them properly in context — but here's the cast of characters so nothing surprises you:
| Term | What it means |
|---|---|
| Workbench | The operational domain you are working in: Conflict, Health, or Disaster. Switching it reconfigures the whole interface. |
| Battlespace | A scoped workspace for one situation (a conflict, region, or investigation) containing the relevant intel, actors, events, and reports. |
| Intel item | A single piece of source material: a report, article, post, image, or geolocated observation. |
| Actor | A person, unit, organisation, or asset tracked across many intel items over time. |
| Event | A confirmed occurrence with a time and place, usually built from several corroborating intel items. |
| Credibility / validity | The weight the platform assigns an item or source, derived from several factors and refined by your decisions. |
| Module | A reusable analytics block — network graph, hotspot map, timeline — you drop onto a dashboard or into a report. |
The same concepts carry across domains. A Battlespace in the Conflict workbench works exactly the same way as an Outbreak in the Health workbench. One is a container for a security situation. The other is a container for an epidemiological one. The system underneath is identical. Once you have learned an idea in one workbench, you have learned it everywhere.
Your first session
A good way to get oriented in about fifteen minutes:
- Read this article, then A Tour of the Workstation to learn the layout.
- Open a battlespace (or create one) so the firehose narrows to a single situation — see Battlespaces.
- Watch the intel feed land scored items, and open one to see how its credibility was derived — see Credibility & Validity.
- Escalate a corroborated item to a confirmed event and find it on the map.
That loop (focus, read, decide, escalate) is the core of daily work. Everything else in the platform builds on it.
How to use this knowledge base
- Browse by category in the left sidebar; each category opens with an overview article.
- Use the search field at the top of the sidebar to jump straight to a topic.
- Inside the app, click the ? button in the bottom-right corner to open the help panel. It searches this same content without leaving your work, and often pre-seeds a query relevant to the panel you have open.
- Press Ctrl+K (or Cmd+K) to open the command palette and search for entities, panels, or commands from anywhere in the workstation.
New here? Read this article, then jump to the category that matches your first task. The docs are reference material, not a manual to read in order.
Where to next
- Creating an Account & Signing In: getting in and staying in.
- A Tour of the Workstation: the layout, named region by region.
- Choosing a Workbench: pick the right operational domain first.