COMPANY BRAIN
Menu
COMPANY BRAINKNOWLEDGE GOVERNANCE FOR AI AGENTS

Govern what
your agents know.

Others govern what agents can do. Company Brain makes sure the knowledge behind every action is current, approved and defensible.

See how it works Technical-fit exploration
BUILT IN JOHANNESBURGSELF-HOST OR CLOUDDESIGN CUSTOMERS WELCOME
LAUNCH PROGRAMME / 2026US$20K first 12-month term→
COMPANY BRAIN / LIVE CONTROL● ANSWER LIVE
GOVERNING SOURCEREFUND
AUTHORITY

v18.1 · Finance Operations

$750manager approval
AGENT ANSWER

Can I approve this refund?

YES / $750Verified against v18.1
CONTROL ROUTE

Source, owner and answer state stay connected.

WATCH
01 / TRY IT
THE SIMPLE PROMISE

When the source changes, every agent updates—or stops.

01ONE APPROVED SOURCE
02NO STALE ANSWERS
03PROOF ON DEMAND
KNOWLEDGE / IN MOTION

Human judgment.
Machine speed.

Company Brain keeps authority visible while trusted knowledge moves through every agent.

A stylised human figure interacting with a layered knowledge coreCB / 001 VERIFIED MOTION
01

THE GAP

THE MISSING HALF OF AGENT GOVERNANCE

Permission is not
truth.

An agent can be perfectly authorised and still act on the wrong policy. Company Brain closes that gap.

ONE AGENT / ONE DECISION“Can I approve this refund?”REQUEST: $750
CHECK 01 / ACCESSYOUR EXISTING CONTROLS

Is the agent allowed to act?

Access controls answer who can use which tool—and within what limits.

Identity verifiedTool access allowedAction scope in bounds
CHECK 02 / KNOWLEDGECOMPANY BRAIN

Is the answer safe to use?

Knowledge controls answer whether the guidance is current, owned and released for use.

Source v18.2 currentOwner approvedAnswer impact released
ONLY WHEN BOTH ARE TRUEAccountable AI.

The agent has permission—and a current, approved answer.

02 / HOW IT WORKS

Three moves.
One trusted answer.

A physical flow from governed source documents, through human approval, to agent-ready knowledge
GOVERNED FLOWSOURCE → HUMAN → AGENT
01

Connect the source.

Bring policies, procedures and operational knowledge into one governed layer.

02

Approve what is true.

A named person reviews the knowledge before any agent can use it.

03

Stay current.

When a source changes, affected answers pause until they are approved again.

03 / CHANGE CONTROL

See the moment an answer
becomes unsafe.

A policy update should not become a scavenger hunt. Compare the source, approval and agent outcome in one control view.

INTERACTIVE DEMONSTRATION
CONTROL DELTA / REFUND AUTHORITYANSWER LIVE
PREVIOUS SOURCEv18.1

Manager approval required over

$300Finance Operations / approved
GOVERNED SOURCEv18.2

Manager approval required over

$750Prepared for named owner review
AGENT ANSWERCan I approve a $500 refund?
LIVENo — manager approval required
RECEIPTSOURCE → OWNER → ANSWER IMPACTVersion linked
04 / EARLY COLLABORATION CONTROLLED BETA / 2026
COMPANY
BRAIN
Raptor Codex logoRAPTOR MCPFOR QLIK

From governed knowledge
to useful insight.

Company Brain and Raptor MCP for Qlik are exploring how governed knowledge can travel safely into AI-assisted analytics workflows.

IN THE ROOM

Demo data / controlled environments / selected design partners

Discuss the controlled beta Explore Raptor MCP for Qlik
A blue governed knowledge archive connected through a controlled signal bridge to a graphite analytics engine
CONTROLLED EXCHANGEKNOWLEDGE → INSIGHT
05 / LAUNCH PROGRAMME 2026

Production proof.
Built in.

Company Brain is opening its first three production customers at a defined Launch Programme rate. This is not an open-ended beta: every customer gets a bounded production environment, a written go-live record and a measurable control route.

The US$20,000 first-year rate is available for order forms signed by 30 September 2026. It includes implementation, enablement and the evidence needed to show what changed.

LAUNCH PROGRAMME / ROUTEREAL PROGRAMME STATUS
OPEN / REVIEWED 2 August 2026Three production places are available.

The Launch Programme is limited to the first three production customers. It is a defined programme, not an unbounded beta.

First-year rate: US$20,000, paid in advance.
Request the Launch Programme scope
PILOT SIGNAL / REVIEWED 2 August 2026 COHORT OPEN
FIRST-YEAR RATEUS$20K12-month term, paid in advance
LAUNCH COHORT03first three production customers
PROGRAMME CLOCK30 SEPsigned by 30 September 2026
01 / DEPLOY01

One real
environment.

Deployment into one customer-hosted production environment, with three standard connectors enabled and an initial corpus processed.

1 ENVIRONMENT / 3 CONNECTORS / 50K DPU BACKFILL
02 / GOVERN02

Publish only
what is owned.

Configure up to three governed domains and 40 production workflows, with named reviewers in the publication route.

3 DOMAINS / 40 WORKFLOWS / HUMAN REVIEW
03 / EVIDENCE03

Leave with
a record.

Track source changes, reviewer decisions, drift alerts and monthly usage so the operating story is evidence-led from go-live.

DRIFT SIGNALS / REVIEW RECORD / USAGE STATEMENT
CUSTOMER-HOSTED / NO PER-SEAT FEESCOPED → OBSERVABLE → PRODUCTION-SCOPED
LAUNCH PROGRAMME / RATE CABINET

US$20,000 for the first 12-month term. Clear expansion rates where published; quote-only items stay quote-only.

GOVERNED CAPACITY
Domain pack
US$6,000 / year
Workflow pack
US$3,000 / year
DPU pack
US$150 / month
Backfill pack
US$750 / 10K DPU
Domain pack: 1 domain + 10 workflows + 1K DPU/month. Workflow pack: 10 workflows. Annual DPU commitment: US$1,500 / year per extra 1K DPU/month.
CONNECTIVITY
Supported connector
US$500 once-off
Additional account
US$300 once-off
API / OAuth build
from US$7,500
Complex / on-prem build
from US$15,000
Three supported standard connectors are included in the Launch Programme. Manual file upload is included.
ADVANCED OPTIONS
Premium model, client key
US$1,000 setup
Managed premium model
provider cost +15%
Operator Mode / external writes
separate quote
Extra environments / 24×7 SLA
separate quote
Managed premium model access has a US$250/month minimum. Operator Mode is quoted only after production-readiness and safety approval.

Published rates are indicative, exclusive of VAT and similar taxes, and subject to the final signed order form. The signed order form and contract control if they differ. The Launch Programme rate applies only to the first three production customers whose order forms are signed on or before 30 September 2026, in exchange for the participation commitments recorded in the order form.

05B / BUYER COMPARISON

Know what you are
actually buying.

Company Brain is not trying to out-search Glean, out-automate Moveworks or replace an AI-governance suite. It makes the knowledge behind high-stakes AI answers current, owned and reviewable—inside a defined production scope.

15 decision factors / scroll sideways to compare
Decision factorCompany BrainGoverned knowledgeGleanEnterprise searchMoveworksAgent automationOneTrustAI governance suite
Primary jobWhat the platform is principally designed to govern.Knowledge behind agent answersEnterprise search and work AIEmployee automation and AI agentsEnterprise AI risk and compliance
Named review before publicationA person can approve, reject or annotate the knowledge used in production.Included: named reviewer routeHuman oversight supported; not its core publication modelApprovals around agent workflowsStrong approval and attestation workflows
Source and version trailCan a buyer trace a governed answer back to the source material and its version?Included: source, version and evidence traceGrounded answers and source permissionsLogs and workflow-level audit trackingSystem, model and control documentation
Source-change controlHow a changed policy or procedure is handled before it becomes a stale answer.Included: drift / contradiction review and dispositionUp-to-date retrieval and connector maintenance focusAgent and policy validation focusAI posture and policy monitoring focus
Evidence for a specific answerEvidence that links the answer, source, owner and change state together.Core design pointCitations and audit activity availableAudit trail for agent actionsCompliance evidence across AI systems
External write actionsWhether the product is designed to execute actions in other systems.Gated separately; not promised in the launch packageAgents can act with scoped, logged permissionsCore strength: agentic automationPolicy enforcement and workflow automation
Enterprise searchBroad discovery across a large app estate.Purposeful, not the primary propositionCore strength: broad public connector catalogueStrong employee-assistant searchNot the primary proposition
Connector breadthOut-of-the-box integration catalogue versus a focused launch scope.3 included; 7 supported standard connectors in launch scopeCore strength: broad native and custom connector estateCore strength: broad integrations and agent marketplaceBroad enterprise ecosystem by module
AI portfolio governanceInventory, risk management and regulatory controls across many AI systems.Focused on governed knowledge for agentsNot its primary propositionAgent-platform governanceCore strength: enterprise-wide AI governance
Customer-hosted environmentA production environment deployed into infrastructure the customer provides.Included: one isolated customer-hosted environmentEvaluate deployment model with vendorEvaluate deployment model with vendorEvaluate deployment model with vendor
MCP and REST accessIntegration into the customer or partner interface instead of a forced front end.Included within launch allowancesAPIs and agent platform availableMCP, plugins and agent integrations availablePlatform APIs and integrations available
Launch scopeWhat is made concrete in the commercial offer rather than left to a discovery call.1 environment, 3 domains, 40 workflows, 50K DPU backfillEnterprise scope is sales-ledEnterprise scope is sales-ledModule and enterprise scope is sales-led
Pricing basisWhether cost grows primarily with users, platform modules or governed operating scope.No per-seat fee; scope, volume and workflowsSales-led; deployment and user scope matterSales-led; user / deployment scope mattersSales-led; modules and enterprise scope matter
First-year commercial anchorA visible starting point for a bounded production programme.US$20K / 12 months for first three production customersContact salesContact salesContact sales
Usage protectionHow customers see and agree to scope expansion.Usage statement; no retrospective surprise overagesConfirm contract terms with vendorConfirm contract terms with vendorConfirm contract terms with vendor
THE BUYER TEST

If the risk is a policy-heavy AI answer that has become stale, unowned or impossible to defend, Company Brain is the focused control layer.

Map your decision route

Comparison based on the Company Brain Launch Package and public vendor materials reviewed August 2026. It is a product-fit guide, not an exhaustive benchmark; validate technical, commercial and security requirements in your own procurement process. Public references: Glean, Moveworks and OneTrust.

05A / WHAT IT CHANGES

Built for the moment
“probably” becomes risk.

01LIABILITY CONTROL
BLOCK

Stop the wrong answer.

When a policy changes, stale knowledge is withdrawn before it becomes customer-facing liability.

02AUDIT EVIDENCE
PROVE

Prove every answer.

Export a clean evidence trail showing what was used, where it came from and who approved it.

03SKILL SUPPLY CHAIN
ALLOW

Trust every skill.

Ground, review and approve agent skills before they enter the organisation’s operating environment.

VISUAL ESSAY / THREE CONDITIONS OF TRUST

Trust needs a
physical language.

Evidence, human authority and change control should be visible—not buried in a system log.

A monumental archive built from interlocking document slabs
01 / EVIDENCE

Evidence has structure.

Auditability should feel built in—not assembled under pressure.

A verified blue policy layer held steady while outdated layers drift away
02 / CHANGE

Drift should be visible.

When the source moves, trust must move with it.

Sculptural hands passing a policy through a human approval point
03 / AUTHORITY

Authority stays human.

THE CONTROL CHAINSOURCE
→ HUMAN
→ AGENT

Every trusted answer keeps its origin and owner attached.

COMPANY BRAIN / 2026
06 / WHERE IT FITSCHANNEL STRATEGY

One control.
Three buyer routes.

Company Brain turns knowledge governance into a practical risk, assurance and evidence proposition—starting with the question each buyer already needs answered.

A blue knowledge core connecting through three verified routes
ONE CONTROL / MULTIPLE ROUTES

ONE KNOWLEDGE CONTROL
THREE CLEAR WAYS IN

01AI-LIABILITY INSURERS

A claims-prevention control.

Give AI liability insurance providers and AI-deploying insureds a practical control for stale, fabricated and unapproved answers - with evidence that supports underwriting conversations.

START WITH A MATERIAL DECISION“Which AI answer could create the most avoidable loss when a policy changes?”

FIRST PROOFA visible stale-answer control and reviewable loss-prevention record.

02AUDIT & ADVISORY

Evidence with a chain of custody.

Give assurance teams a reviewable record of the source, approver and change state behind every governed answer.

START WITH ONE TESTABLE ASSERTION“Can the team trace this agent answer back to a source and accountable owner?”

FIRST PROOFAn Evidence Pack that shows origin, authority, change and impact together.

03EU AI ACT / GRC

Traceability that is ready to review.

Support EU AI Act compliance readiness with a reliable evidence layer for governance, record-keeping and internal-control workflows as obligations evolve.

START WITH ONE HIGH-IMPACT USE CASE“Where would a reviewer need proof that an AI answer was current and governed?”

FIRST PROOFA traceability workflow to support internal governance and readiness conversations.

Company Brain supports governance, evidence and compliance-readiness work; it does not itself provide a legal certification or replace professional assurance.

07 / SECTOR ROUTESSELECT A WORKFLOW → TRY THE CONTROL

See Company Brain
in your world.

Choose your environment—global insurer, specialist firm or growing team—then change the source. The demonstration shows the moment an AI answer becomes unsafe and what a governed route does next.

INSURANCE & CLAIMS / LIVE SCENARIOCURRENT
01 / WHERE IT FITS

Prevent policy drift from becoming claim exposure.

GOVERNING SOURCEClaims authority policyv18.1 · Owner: Head of Claims
Over $300
AGENT ANSWER

Does this claim need manager approval?

Over $300Verified against v18.1
CONTROL EFFECT

The answer is live because the source and owner are current.

FOR

Claims leaders · underwriting teams · broker operations

START WITH

Claims authority, cover interpretation or broker guidance that changes often.

FIRST EVIDENCE

A dated policy version, claims owner approval and affected guidance record.

1 / Change a material source
08 / PRODUCT PROOF

Every answer
has a receipt.

An Evidence Pack keeps the source, the person accountable for it, the change event and the affected agent answer in one reviewable record.

It gives assurance teams something more useful than a claim: a clear chain they can inspect, export and discuss with the control owner.

01 / ORIGINSource-linked

The policy and its version stay attached to the answer.

02 / AUTHORITYNamed owner

One accountable reviewer decides what is true.

03 / CHANGEImpact visible

Affected answers pause until the change is reviewed.

Built to support evidence and traceability workflows; it does not itself certify legal or regulatory compliance.

EVIDENCE PACK / CB-024 REVIEWABLE
GOVERNING SOURCEPOLICY / REFUNDS
v18.2

Updated 14 May 2026 · Finance Operations

✓
HUMAN APPROVALPriya N. / Head of Claims

Approved for agent use · 14:42

CHANGE EVENT$300 → $750

Approval limit revised in source policy.

AGENT IMPACT2 ANSWERS PAUSED
ILLUSTRATIVE CONTROL RECORDORIGIN → OWNER → IMPACT
CONTROL ROUTE / 03 READY FOR REVIEW
01 / CONNECTSource
systems
POLDOCKB
Policies · procedures · knowledge bases
02 / GOVERNReview
queueOwner, version and approval stay attached
03 / RELEASEAgent-ready
knowledge
AI

Which refund limit applies?

$750 VERIFIED
WORK WITH THE STACK YOU ALREADY CONTROLSOURCE → REVIEW → ANSWER
09 / THE OPERATING LAYER

Keep your stack.
Govern the answer.

Company Brain is designed to sit alongside the identity, access and runtime controls you already use. It gives source knowledge a clear route to an agent: connected, reviewed, versioned and released with an owner.

YOUR SOURCESBring the policies and operational knowledge that already run the business.
YOUR CONTROL OWNERSMake the reviewer, version and approval decision visible before release.
YOUR AGENT LAYERGive agents a governed answer without replacing the systems that make them useful.
Map a workflow
10 / VALUE MODEL

Make control
pay for itself.

This model follows two costs Company Brain is built to reduce: rechecking answers that lack a clear source, and fixing answers after a governed source changes.

Planning model only. It is not a savings promise, quote or claim of guaranteed outcomes.

COMPANY BRAIN / VALUE ESTIMATOREDIT YOUR OPERATING REALITY

ASSUMPTIONS YOU CONTROL

ILLUSTRATIVE ANNUAL CAPACITY
R 576 000value of time redirected from avoidable control work
LESS MANUAL CHECKING720 hrs
LESS STALE-ANSWER REMEDIATION240 hrs
TOTAL TIME REDIRECTED960 hrs

Company Brain connects the source, accountable owner and answer impact. Start with the one workflow where your team currently spends the most time checking whether an answer is still safe.

Map this to a workflow →
11 / FIELD NOTES

Three ideas
behind the build.

FN—01HUMAN AUTHORITY

Confidence is not approval.

CONFIDENTis not the same as correctCB
FN—02DRIFT

A policy is only current until it changes.

v17→18trust must move tooCB
FN—03CONSISTENCY

One source. Every agent.

1→∞approved truth, reusedCB
RESEARCH / POPIASSRN

Selling
Ourselves

POPIA & SYNTHETIC DATA MARKETS

12 / RESEARCH

Who remains accountable when data changes form?

Founder Kai Swarts' research examines consent, value and responsibility in synthetic-data markets—the same questions of provenance and accountability that shape Company Brain.

CONSENTVALUERESPONSIBILITY
Read the paper on SSRN
13 / TRUST, WITHOUT THE THEATRE

The questions
that should be asked.

High-trust AI needs more than a good interface. It needs a clear answer to who controls the knowledge, what changes and what can be proved.

VISIBLE ORIGINNAMED AUTHORITYCHANGE AWARE
BUYER DUE DILIGENCE / 05 ANSWERED DIRECTLY
01Is Company Brain another chatbot?

No. It is a knowledge-governance layer for the material an agent relies on: where it came from, who approved it, what changed and whether it is still safe to use.

02Does it replace our identity, access or agent-runtime controls?

No. Those controls remain important. Company Brain is designed to complement them by governing the knowledge behind an agent’s answer, not by replacing the systems that control what an agent can do.

03Does this make an organisation EU AI Act compliant?

No product can make an organisation compliant on its own. Company Brain can support governance, record-keeping, traceability and internal evidence workflows as part of a wider compliance-readiness programme.

04Who decides what an agent is allowed to know?

A named human owner does. Company Brain is built around visible authority: the source, reviewer and approval decision stay attached to the knowledge route.

05What does a first engagement look like?

A controlled, scoped workflow: choose one material decision, use an appropriate bounded environment and finish with an evidence-led view of the control gap and next decision.

14 / FOUNDER
JOHANNESBURG
SOUTH AFRICA

KAI SWARTS / FOUNDER & BUILDER

Company Brain was built for accountable AI.

Kai works at the intersection of law, systems and AI accountability. He founded Company Brain on a practical conviction: agents should never be more certain than the knowledge beneath them.

COMPANY BRAINAI GOVERNANCEPOPIA RESEARCHBCom LAW · UP
15 / OPEN CHANNEL DESIGN PARTNERS

Bring the workflow
that cannot afford “probably.”

If your agents answer policy-heavy questions, operate in a regulated environment or need a trusted skill supply chain, Company Brain wants to understand what breaks today.

Start a founder-led scope