Join the waitlist
Menu

KAIROS DYNAMICS

AI can generate. The bottleneck is deciding what deserves to advance.

Kairos Dynamics productizes a constraint layer for delegating consequential work to software. Evidence, context, coordination, and authority remain explicit rather than collapsing into one opaque agent.

Evidence must be earned.Context stays scoped.Authority stays bounded.

Invariant is the first commercial wedge. Financial markets are the first proving domain because weak models fail there quickly and measurably.

Inspect the evidence record →See the architecture
QUALIFICATION FIELD / WHAT YOU ARE LOOKING ATEach trace is one possible predictive relationship.

Read left to right: candidate models or decision rules face six evidence tests before forward observation and a separate authority boundary.

Research layer
Implemented Heatmap Strategy Lab ancestry
Advancement
Kairos Core / specified
Playback
Deterministic schematic
SEARCH / GENERATION 0056 proposed · 0 rejected · 0 observing · 0 advanced
Resolve the instrument
candidate traceruled-out memoryunderpowered / retestforward observationcontrolled advancement

A conceptual preview, not live results. The observatory explains every gate, outcome, and evidence boundary.

PRODUCT SYSTEM

Four control problems. Four systems.

Each system governs a different state transition. Invariant handles evidence, Context Passport handles portable context, Valence handles privacy-preserving coordination, and Lightcone handles authority.

  1. Invariant

    EvidenceOPERATING

    Has this candidate earned the right to advance?

    The first commercial wedge. A customer supplies a research mandate, data, universe, risk limits and privacy boundary. Invariant performs recurring generation, challenge, preservation, qualification and forward monitoring, returning a qualified candidate stream with evidence, lineage and advancement state attached.

    Inspect the record →

  2. Context Passport / PHC

    ContextSPECIFIED

    Can context travel without surrendering control of the person it describes?

    A user-controlled context architecture for carrying scoped, provenance-aware information across AI hosts. The format is intended to remain portable and separable from permission to act; commercial value, where created, sits in implementations and management tooling rather than ownership of a person's context.

  3. Valence

    CoordinationPRIVATE PREVIEW

    Can software surface a legitimate route to a goal without exposing more context than the route requires?

    Privacy-preserving coordination and intent routing with canonical identity, consent and local/private deployment options. Valence is the coordination system in the Kairos architecture; it is not evidence qualification and it does not inherit authority merely because it can identify a route.

  4. Lightcone

    AuthorityIMPLEMENTED

    Was this action allowed, and can you prove it?

    A control boundary for autonomous systems: authority delegated in bounded envelopes that fail closed, consequential actions checked before they leave, and decisions recorded with receipts. It is implemented and tested; the site does not represent it as a production customer deployment.

    Inspect the record →

One architecture, different workflows. Kairos Core and Domain Packs sit underneath this public taxonomy as implementation layers. They are not additional control problems or competing product families. See maturity and boundaries →

FIRST COMMERCIAL WEDGE / INVARIANT

Delegate the research mandate, not just the tooling.

The customer supplies the mandate, data, universe, risk limits, and privacy boundary. Invariant performs the recurring research work and returns a qualified candidate stream with evidence, lineage, and advancement state attached.

  1. 01Generate

    Create candidate strategies or models inside the customer's mandate and data boundary.

  2. 02Challenge

    Attack candidates with falsification, holdouts, costs, controls, and regime tests.

  3. 03Preserve

    Keep failures and nulls so the search does not rediscover its own dead ends.

  4. 04Qualify

    Advance only candidates whose evidence state meets the bar fixed before the decision.

  5. 05Monitor

    Keep forward behavior, lineage, and advancement state attached over time.

First buyer hypothesis: systematic funds, prop firms, and quant teams that want more research throughput without surrendering evidence discipline.

WHAT YOU JUST WATCHED

What must a prediction survive before it is allowed to influence a decision?

The field above runs fifty-six of these at once. Here are three, slowly.

  1. Candidate ALooks convincing on the history it was found in.

    REJECTEDIt never separated from noise. The rejection is kept.

  2. Candidate BGets further. Survives the first challenges.

    MORE EVIDENCE REQUIREDNot refuted, but not established either. It waits rather than advancing.

  3. Candidate CClears every historical gate.

    FORWARD OBSERVATIONNow it has to survive time it has never seen.

Evidence does not grant authority.

Candidate C has earned the right to be watched, not the right to act. Crossing that boundary is a separate, explicit authorization step, and it can be withdrawn without the evidence changing at all.

Three schematic candidates, drawn to explain the gate sequence. Not a recorded HSL result, not a performance claim, and not Kairos telemetry. The gate order and names are the ones the research system uses.

Open the Observatory
RECORDED RESEARCH ANCESTRY / HSL

The first wedge is backed by an operating proving environment.

Heatmap Strategy Lab is research ancestry, not a fifth Kairos product. It supplies a record of generation, rejection, failure memory, qualification, and forward observation that can be inspected without presenting research evidence as customer traction or live-capital performance.

RECORD / 8,855 typed rejection records against 374 archive occupants. 59.9% of those occupants are under a contamination notice.

RECORDED TRAINING-GYM REJECTIONS

Why do candidates stop?

A training gym is one of the search programs XGCS leaves running: it proposes candidate predictions continuously, and a referee decides which are allowed to enrol. The field above is a schematic. This is the real register, every rejection those gyms recorded across the4 of 8 programs that have logged any, sorted by the reason each was attributed to.

92.4%

of 8,855 recorded rejection records stopped at one place:never separated from noise. Almost nothing survives far enough to fail for an interesting reason.

  1. 8,186never separated from noise92.4% of the register · V1 / holdout

The other 669, shown at their own scale. Together they are 7.6% of the register.

  1. 317a simpler baseline already explained itR2 / spanning
  2. 209scrambled data scored just as wellR1 / placebo
  3. 101failure mode not resolved by the export classifierunclassified
  4. 42did not survive trading costsR4 / net of cost

What this shows. Where the search's own rejections were attributed, across every program that has recorded any.Why it matters. The rejections are kept rather than discarded, so the register of what did not work is itself part of the research record.

BOUNDARY / This is implemented XGCS research ancestry, not Kairos customer telemetry or historical performance. Kairos authority and deployment states remain a separate specified architecture.

Recorded process activity is not proof of predictive quality, return, confidence, or deployment authority.

AS OF Aug 26, 2026, 10:19 PM UTC

Recorded activity from the XGCS training gym, which is the same record on a time axis and can be scrubbed frame by frame. Sanitized for publication. Counts are rejections attributed to a failure class, not a survival funnel: the referee wires its checks into one enrol-or-reject decision rather than a sequence of stages, and it qualifies candidates rather than ending them. A check appears here only when the export classifier resolved rejections to it. The unclassified row is known to contain persistence failures that the classifier did not resolve, so the absence of a row is not evidence that a check recorded nothing.

hsl_20260826T221952Z_30f6d81ac86e
BOUNDARY / WHAT THIS PROVES

Working machinery exists. Commercial validation still has to be earned.

OPERATING RESEARCH ANCESTRY

Invariant + HSL

The research loop runs today in the founders' environment. Productizing it into repeatable delegated customer work is the first commercial objective.

SPECIFIED

Context Passport / PHC

Portable, purpose-scoped context with provenance and a boundary between knowledge and permission.

PRIVATE PREVIEW

Valence

Privacy-preserving coordination, consent, and local/private deployment patterns.

IMPLEMENTED

Lightcone

Authority controls, escalation, receipts, and outcomes. No production customer deployment is claimed.

BOUNDARY / These systems do not prove enterprise product-market fit, verified live trading alpha, a finished cross-domain product, or customer deployment across the portfolio.

DEPLOYMENT PRINCIPLE

Software should go to the data.

01 / PREFERREDCustomer-controlled private

Run Kairos where sensitive data already lives, under customer security and governance rules.

02 / APPROPRIATEPrivate cloud or VPC

Keep strategy IP and organizational context inside a customer-approved trust boundary.

03 / WHERE SUITABLEManaged deployment

Use hosted operation only where confidentiality, latency, and policy requirements allow it.