Skip to content

Raven

Intelligence. Research. Engineered Systems.

An advanced proprietary technology ecosystem developed by ResortOS, bringing together quantitative research, intelligent analysis, automated systems and coordinated software architecture.

Developed by ResortOS

One system, built from parts that each do one job

Layers
Six, from the data foundation up to coordination
Components
8 services and applications, each with one job
Hosting
Self-hosted, including its language model
How it changes
Preregistered research, then small versioned releases

Raven is a proprietary software ecosystem developed by ResortOS. It brings data collection, quantitative research, model-assisted analysis, automated agents, monitoring and coordination together in one system that the team runs and develops itself.

Its purpose is simple to state and hard to do well: turn large amounts of changing data into careful, explainable, automated work, and keep people in charge of it. Every part exists to make that work more reliable, better tested or easier to see.

Each part is a separate service with a clear responsibility and a defined way of talking to the others. They share one data foundation, one coordinator and one set of limits. Because the parts are independent, any one of them can be stopped, replaced or improved without taking the rest down.

Raven is built around research. A change starts as a written, testable idea. It is checked against history and against chance, then run in shadow beside the live system, and it is adopted only if the evidence holds. Most ideas don’t pass, and the record of why is kept.

The system is designed to keep evolving in small, reversible steps. New parts begin by observing and recording before they are allowed to act, every release leaves a rollback point, and decisions and incidents are written down so each change starts from what was learned last time.

A raven’s feathers are black. Their blue and green sheen comes from the structure of the feather, not from any pigment.
Why the name: in a well-built system, the useful behaviour comes from the structure too.

Six layers, one direction of trust

Data enters at the foundation and is checked before anything uses it. Research decides what is allowed to change. Coordination at the top can narrow what happens below it, never widen it. Choose a layer to see what it does and what it shares with the others.

Sends to Receives fromConceptual view. Not a map of real services or connections.

Data Intelligence

Collects, checks and stores the data everything else relies on.

External time-series data is collected continuously, converted into one consistent format, checked against its original source, and kept in a tiered cache built to recover cleanly after a restart or a power cut. Every other layer reads from here.

Sends

  • Historical data, to
  • Current context, to
  • Current data, to
  • Freshness checks, to

Receives

Nothing directly.

The components

Eight parts of one internal system. Each belongs to a layer of the architecture and works with a small, known set of neighbours. They are not sold as separate products.

  • Foundation

    Raven Core

    The shared foundation every other component is built on.

    The shared message bus, durable storage and the data hub with its crash-safe cache. Core collects data once and serves the same picture to every component, so no two parts ever work from different versions of it.

    Layer
    Data Intelligence
    Works with
    Raven Research, Raven Intelligence, Raven Automation, Raven Risk
  • Evidence

    Raven Research

    Where ideas are tested before they are used.

    Historical replay, tests that are written down before they are run, random baselines and forward shadow runs, with a permanent record of every result.

    Layer
    Quantitative Research
    Works with
    Raven Core, Raven Automation, Raven Control
  • Explanation

    Raven Intelligence

    Locally hosted models that explain what the system sees.

    A model service with a shared priority queue, strict response formats and ongoing measurement of whether its output is actually useful.

    Layer
    Analytical Intelligence
    Works with
    Raven Core, Raven Automation, Raven Control
  • Model training

    Raven Academy

    Where the locally hosted model is trained and examined.

    Fine-tuning and written examinations for the language model. Every candidate model is scored before it is considered, and most are not adopted.

    Layer
    Analytical Intelligence
    Works with
    Raven Intelligence, Raven Research
  • Execution

    Raven Automation

    The agents that carry out defined work.

    Independent services, each with its own rules, state and log, designed to resume exactly where they stopped.

    Layer
    Automation Systems
    Works with
    Raven Core, Raven Risk, Raven Control
  • Protection

    Raven Risk

    Guards, checks and alarms.

    Circuit breakers, health checks, heartbeats and a daily audit that compares the system's own records with their original sources.

    Layer
    Risk and Monitoring
    Works with
    Raven Core, Raven Automation, Raven Control
  • Coordination

    Raven Control

    Coordination and oversight in one place.

    The coordinator that assigns work and settles conflicts, and the control application where people see the whole system and approve sensitive actions.

    Layer
    Orchestration and Control
    Works with
    Raven Automation, Raven Risk, Raven Research, Raven Staff
  • People

    Raven Staff

    A workspace for supervised, person-led work.

    Tools for carrying out work by hand alongside the automated agents, with biometric approval on a phone for sensitive steps.

    Layer
    Orchestration and Control
    Works with
    Raven Control, Raven Risk

The bars beside each role show where the component sits among the 6 layers, top to bottom.

How Raven is engineered

The interesting part of Raven isn’t any single component. It’s the discipline around them: how ideas are tested, how independent agents are kept in step, and how the whole system stays visible.

  • Research before change

    Every idea is written down as a test, replayed against history, compared with chance and run in shadow before it can change anything.

    How research works
  • Coordinated automation

    Independent agents share one coordinator, one data foundation and one set of limits, and people approve the actions that matter.

    How agents are coordinated
  • Visible operation

    Heartbeats, a daily audit, alerts on the team's phones and releases that always leave a way back.

    How Raven is monitored

ResortOS

Developed by ResortOS

Raven is a proprietary technology ecosystem developed by ResortOS, reflecting our work in advanced software engineering, quantitative research, intelligent data systems and automation architecture.

ResortOS builds software products, AI systems and client projects. Raven is the company’s own long-running engineering project: the place where its research, data and automation work is built, tested and run every day.

Talk to us about Raven

For questions about Raven, or about working with ResortOS on research, data or automation software, write to the ResortOS team.

Raven is used inside ResortOS. It isn’t offered as a public service, so this site can’t set up accounts or access.