Early access · Relara v0

Governed context
for AI agents

Relara turns private emails, meetings, docs, tickets, and repos into approved artifacts your agents can use, cite, and publish safely.

relara · proposal #PR-0421
Pending review
Private sourcewithheld
█████ ██████ ████ ████████
Sensitive · contains PII
evidence
Derived artifact
Top 5 churn signals + evidence spans
Approved snippets · owners attached
evidence
Policy decision
Share to project:onboarding
Redact: customer names · enterprise tier only
evidence
Context grant
agent:retention-copilot
Read-only · expires in 30 days
Evidence drawer4 spans · 3 calls
“…we ended up renewing only because support was responsive.”
0.92
Call · 09/14 · 02:41view span →
“The setup took two weeks longer than we expected.”
0.87
Call · 09/22 · 18:09view span →
“If onboarding had a checklist we'd have shipped sooner.”
0.81
Call · 10/03 · 11:55view span →
Policy checks
PII redaction
Tier scope
Owner approval
!External share
Receiptrcpt_0421_a91
Approver
dana@relara
Destination
agent:retention-copilot
Redactions
3 spans · names
Revocable
yes · 30d
signed · ed25519
policy v2.3

Trusted by fast-growing teams building with agents

NorthwindLATTICEvercel▲Helix.aiPARALLELstellar/QuantaOPENGRIDkepler.coArcadiaNorthwindLATTICEvercel▲Helix.aiPARALLELstellar/QuantaOPENGRIDkepler.coArcadia
The problem

Your agents need context. Your company cannot just dump everything into memory.

Useful knowledge lives in inboxes, meeting transcripts, docs, issue trackers, and databases. Agents need that context to help, but raw access creates permission, privacy, and trust problems.

01

Private sources

Email threads, transcripts, and customer notes often contain useful signal mixed with sensitive details.

02

Messy permissions

Search and RAG systems often lose the difference between who can know something and who can cite it.

03

Unsafe agent action

Agents should not publish, write back, or brief a team from unreviewed private context.

How it works

From private source to approved agent context

  1. step 1

    Connect sources

    Bring in emails, meetings, docs, GitHub, Linear, Slack, Drive, Notion, databases, or local files.

  2. step 2

    Analyze privately

    Relara extracts decisions, summaries, claims, tasks, evidence, and graph links without making raw sources broadly visible.

  3. step 3

    Review proposals

    A proposal shows what was found, why it matters, what stays private, and exactly what approval would share.

  4. step 4

    Grant context

    Approved artifacts become policy-scoped context grants for agents, projects, teams, or operating loops.

  5. step 5

    Keep receipts

    Every publication or writeback records evidence, approver, destination, redactions, and revocation status.

The governed context layer

Everything agents need to use company knowledge safely

Relara connects your sources, analyzes privately, proposes safe artifacts, and gives agents only the context they are allowed to use.

Governed Context

Turn scattered work data into approved artifacts, graph context, and receipts your agents can trust.

  • Private by default
  • Evidence-backed artifacts
  • Policy-scoped agent context

Source Connectors

Bring in emails, meetings, docs, tickets, repos, Slack threads, and databases without moving your team into a new workspace.

maildocsslackgit

Private Analysis

Extract decisions, claims, tasks, summaries, and evidence while raw sources stay protected.

Policy & Redaction

Apply inherited permissions, sensitivity checks, redactions, and approval rules before anything is shared.

Context Grants

Give each agent, project, team, or operating loop only the approved artifacts it can use and cite.

grant.yamlagent:retention-copilotscope:project:onboardingread:approved-artifactsexpires:30dgranted

Receipts

Record what was approved, who approved it, where it went, and which evidence supported it.

APPROVEDrcpt_a91

Developer SDK

Use Relara as the governed context layer for your own agents with local fixtures, connector manifests, and MCP-compatible tools.

$ relara init→ connectors.yaml created$ relara analyze ./fixtures→ proposal pr_a91 ready$ relara grant agent:demo
Who it is for

Built for teams putting agents into real work

Developer teams

Give coding, product, and support agents approved context from repos, issues, docs, and meetings without exposing everything.

Operators & founders

Turn calls, emails, and customer signals into reviewed project memory and next actions.

AI platform teams

Add policy, provenance, proposals, and receipts to internal agents without building the governance layer from scratch.

For developers

Use Relara as the governed context layer for your agents

The standalone version is planned as a local-first developer package with a CLI, TypeScript SDK, connector manifests, policy fixtures, and MCP-compatible tools.

Start local. Prove the trust loop. Add managed connectors when your team is ready.

~/relara
$ relara dev seed transcript-decisions
→ 3 fixtures loaded · graph built in 240ms
$ relara analyze ./fixtures/transcript.md --target project:demo
→ proposal pr_a91 ready · 5 artifacts, 2 redactions
$ relara proposals approve latest
→ approved · receipt rcpt_a91 written
$ relara context get --target project:demo --agent demo-agent
→ 5 artifacts, evidence-linked, revocable
For companies

Designed for trust before scale

  • Inherited permissions where available.
  • Human approvals before risky sharing.
  • Redaction and sensitivity gates.
  • Evidence-backed artifacts.
  • Context grants for agents and operating loops.
  • Receipts for publication and writeback.

Relara is not another place to move your company knowledge. It is the governed layer between the tools you already use and the agents you want to trust.

Give agents the context they need,
without losing control of the source.

Follow the build →