scalar
Participate
PHRONESIS LABS / DISTRIBUTED INTELLIGENCE
INTELLIGENCE, WITH EVIDENCE.

AI that
shows
its work.

A distributed AI network designed for useful-work mining, independent validation, and Braid provenance.

Explore the architecture Build with us
Pre-launch · Tokenless testnet in development
Useful-work mining/Proof of AI/Braid
THE EVIDENCE ENGINE
PROTOCOL CONCEPT
Scalar evidence engine: mine, validate, and braidRaised compute blocks perform bounded AI work. Independent validation sits on the second layer. Three illuminated evidence strands connect the layers and weave into a Braid provenance receipt. Motion illustrates the proposed protocol, not live network activity.03 / TRACEBraidBraidConnected provenance02 / VERIFYProof of AIProof of AIIndependent validation01 / COMPUTEUseful-work miningMiningUseful AI computationBRAID RECEIPTWork + proof + lineage
SCALAR SAVINGS LAB

What if the next build
didn't start from zero?

Adjust the project. Adjust the network. See what qualified reuse could change.

ILLUSTRATIVE MODEL · NOT A QUOTE · NO SAVINGS GUARANTEED
EXAMPLE WORKLOADS
Project size$100K
History availableMODERATE HISTORY
New networkMature network
Requirement similarityMODERATE
NovelRepeatable
Validation requirementSTANDARD
Traditional estimate$100K
Scalar estimate$89K
Potential difference~$11K

~11% lower modeled production cost · illustrative only

WITHOUT SCALAR
NEW PRODUCTION · $100K
WITH SCALAR
New work
$77K
Adaptation
$6K
Validation
$3K
Network
$3K
Qualified reuse · $23K avoided
WHY DID IT CHANGE?
History found33%Candidate prior capability discovered.
Qualified for reuse23%Candidate capability that survives provenance, rights, quality, security and compatibility checks.
Work still created new77%Requirements that still require fresh production.
Validation carried forward8%Prior evidence judged applicable to the new context.
CANDIDATE REUSE ≠ QUALIFIED REUSE ≠ SAVINGS
NETWORK MATURITYILLUSTRATIVE — Scalar does not currently have this history

More completed jobs → more candidate history → more qualified capability → less duplicated work → narrower validation delta → lower modeled burden.

TODAY
$96K
100K JOBS
$89K
10M JOBS
$85K
SCENE 01THE WASTE

Do the work.Prove the work.Keep its history.Reuse what still holds.

We keep paying
to create what we already created.

  • Code.
  • Research.
  • Tests.
  • Models.
  • Decisions.
  • Evidence.
CODE MODULERESEARCHTEST RESULTSIMULATIONWORKFLOWMODEL OUTPUTEVIDENCEDECISIONCONFIGVALIDATIONDATASETBENCHMARKSPECREVIEW

Most systems keep the output — and lose the history that made it useful.

Scalar is designed so useful work can keep its history.

↓ Follow one job through it

SCENE 02ONE JOB

Do the work.Prove the work.Keep its history.Reuse what still holds.

Watch one job become
something the network can remember.

  1. Perpetuals sends one bounded request.

PERPETUALSAPPLICATIONSCALARCOMPUTE 1COMPUTE 2COMPUTE 3COMPUTE 4VALIDATOR AVALIDATOR BVALIDATOR CCORTEX · PLANNEDBTC LIQUIDATION SCENARIO

Illustrative lifecycle. Not live network traffic.

↓ Its Braid record continues

SCENE 03BRAID

Do the work.Prove the work.Keep its history.Reuse what still holds.

The result is not
the whole story.

Scalar preserves the evidence, work and disagreement behind it.

JOBrequestrequirementsconstraintsWORKexecutionartifactproofVALIDATIONvalidator Avalidator Bvalidator CDISSENT RETAINEDDISPUTE → BRANCHQUARANTINE → TERMINATEBRAID RECORDACCEPTED WITH DISSENT

Agreement can be recorded. Disagreement can be recorded.
Neither is rewritten after the fact.

Illustrative record. Consensus does not imply objective truth. ↓ The record moves into memory
SCENE 04CORTEX · PLANNED

Do the work.Prove the work.Keep its history.Reuse what still holds.

What holds up
does not have to disappear.

CAPABILITYCODEWORKFLOWRESEARCHMODELEVIDENCECONFIGURATIONVALIDATIONBRAID RECORDNEW REQUIREMENTNEW EXECUTION PLAN · REUSE + ADAPT + NEW WORK
Cortex is planned. Reuse only when qualified — savings are never guaranteed. ↓ Across many applications
SCENE 05APPLICATION UNIVERSE

Do the work.Prove the work.Keep its history.Reuse what still holds.

One network.
Many applications.

SCALARTEST HARNESSPerpetualsDerivativesNOT CONNECTEDNeuroTrader RetailTradingTEST HARNESSEventEdgeEvent marketsSIMULATED ADAPTEREventEdge TVMediaPLANNEDPhronesisVenture studioTEST HARNESSNeuroTrader InstitutionalInstitutionalPLANNEDNeuroTrader Capital MarketsCapital marketsNOT CONNECTEDArileus Investor / AureliusInvestor toolsPLANNED

Applications do not share private data by default. No production integration is implied.

Explore the ecosystem
PHRONESIS + SCALARINITIAL DEMAND

The factory
and the memory.

C1C2C3BUILDPROVEREMEMBERREUSEBUILD AGAINLESS REPEATEDWORK EACH CYCLE

Phronesis builds. Scalar proves and remembers. Future builds can inherit what still holds.

PARTICIPATETHREE WAYS IN

A network needs
more than one kind of participant.

BUILDBring work worth doing.
COMPUTELend capacity to real work.
VALIDATEBe the independent check.
ONE PROVENANCE HISTORYIndependent roles. One shared record of who asked, who worked and who checked.
Find your role
ROADMAPNO DATES · EVIDENCE GATES

Decentralization
is a process.

  1. 01Controlled testnetYou are here
  2. 02Qualified participantsUnlocks after evidence + reliability
  3. 03Independent operatorsUnlocks after evidence + reliability
  4. 04Federated networkUnlocks after evidence + reliability
  5. 05Broader decentralizationUnlocks after evidence + reliability
View roadmap

The next job
should know
what the last job learned.

Explore Scalar See how it works
PRE-LAUNCHTOKENLESSIN DEVELOPMENTDisclosures