Your operating intelligence should live where your team can inspect it.
Saberra is built so your team can see the records, review what becomes trusted, and trace answers back to sources. Client data lives in client accounts.
Your tools
Trusted memory path
Trust starts with records your team can inspect.
Saberra does not ask you to trust unsourced AI output. The system creates candidates, humans review them, and Sera answers from the documented record.
- Client data lives in the client's Notion workspace and tool accounts.
- AI extraction creates draft or candidate records for human review.
- Sera answers from reviewed organizational records with source context.
- The standard deployment uses Google Workspace, native Google Meet capture, Notion, an AI provider account, Railway, and a dedicated inbox.
- Zoom, Teams, and other meeting platforms can be captured when transcripts or summaries are emailed into the dedicated capture inbox.
- Notion is the default memory backend. Postgres or additional systems of record can be scoped for larger or more technical deployments.
Where each piece of data goes.
Every step in the pipeline runs inside accounts your organization controls.
| Source | What is captured | How it flows | Who controls the account |
|---|---|---|---|
| Google Meet | Meeting transcript and summary | Forwarded to capture inbox → AI extraction → Notion draft record | Your Google Workspace |
| Email threads | Message content and attachments | Forwarded to capture inbox → AI extraction → Notion draft record | Your Gmail or Google Workspace |
| Zoom / Teams / other platforms | AI-generated summaries or transcripts | Emailed into dedicated capture inbox → same extraction pipeline | Your email account |
| AI extraction layer | Draft decisions, tasks, risks, roles, policies, sources | Written to Notion review queue, not trusted until human approved | Your Notion workspace |
| Sera queries | Natural language questions from your team | Answered from human-reviewed Notion records with source citations | Your Anthropic account (API keys stay in your hands) |
Your data lives in your workspace. Not ours.
Saberra is configured inside accounts your organization controls: Google Workspace, Notion, your AI provider account, Railway, and a dedicated capture inbox. When the engagement ends, every record stays in your workspace. Nothing moves to a vendor database.
Your records stay. No extraction required.
Every approved decision, task, risk, role history, policy, and source record lives in your Notion databases. At any point you can export everything from Notion as CSV, PDF, or Markdown. There is no data that lives only on Saberra infrastructure.
You control the keys. You can revoke access at any time.
The pipeline runs on API keys and OAuth connections your organization administers. To offboard, you revoke access from your Google Workspace settings, your Notion workspace, and your AI provider account. Saberra has no standing access to retain.
Configuration, not content.
Saberra may retain configuration details (deployment settings, integration routing) needed to support active clients. Organizational content, including decisions, meeting records, task history, is never held in Saberra systems.
Security and data ownership.
Does Saberra store our data on its own servers?
No. The standard deployment routes data through your own Google Workspace, Notion workspace, AI provider account, and Railway instance. Saberra configures the pipeline inside accounts you control. Your organizational memory lives in your Notion databases, not in a Saberra database.
What AI provider does Saberra use?
Saberra runs on Anthropic Claude. The extraction and Sera layers run through a Saberra-managed Anthropic account. Your data is never used to train models, and Saberra does not aggregate your data with other clients.
Who can see our Notion records?
Only people with access to your Notion workspace. Saberra's deployment team has no standing access to your workspace. Access is managed through your organization's Notion permission settings.
What happens when the engagement ends?
Every reviewed record stays in your Notion workspace. You own the databases, the content, and the structure. You can export everything from Notion at any time. Saberra holds no copy of your organizational memory.
Can we review what the AI extracted before it becomes organizational memory?
Yes, that is the core of how the system works. AI extraction creates candidates. Nothing becomes a trusted record until a designated reviewer approves, edits, or rejects it in the review queue. Sera answers only from approved records.
What integrations are in the standard deployment?
The standard stack is: Google Workspace (Meet capture and email), Notion (memory backend), Anthropic Claude (AI extraction and Sera), Railway (orchestration), and a dedicated capture inbox. Zoom, Teams, and other meeting platforms work when transcripts or summaries are emailed into the capture inbox.
Processing personal data on behalf of your organization? Review our Data Processing Agreement for the full terms of how Saberra handles data as your processor.
Platform Technical and Governance Documentation
Architecture overview, data residency table, security guardrails, CCOS governance design, and education scope boundary. For data privacy officers and procurement teams.
An AI operations layer should make trust easier to inspect.
Start with the Living Memory Hub demo, then book a focused walkthrough when you want to evaluate fit, tools, source boundaries, and human approval ownership.