Skip to content
[ System of record ] for regulated industries

The agentic
system of record.

AI agents do the work and named people stay accountable. Every record, signature, and decision can be verified offline.

Scroll
00 What it is

Tabellix manages official records, including documents, approvals, signatures, retention requirements, and a tamper-evident history of every change.

Software vendors integrate Tabellix into their products instead of building their own record-keeping infrastructure. It works behind the scenes, giving people and AI agents secure access, consistent permissions, and a verifiable audit trail.

01

Documents and records

Every change creates a new version, and each file is encrypted with its own key.

02

Signatures and workflows

Review and approval workflows with electronic signatures, each bound to its meaning and to the exact version signed.

03

Retention and holds

Retention schedules, legal holds, and approved disposition for every record type.

04

Audit trail

A hash-chained history of every change, and of access at the level you configure, with signed checkpoints that can be verified offline.

Tabellix is named after the tabellions, the public scribes of Roman and medieval Europe who drafted and kept legal deeds.

Records, not files

01 Agentic by design

Agents act. People answer for it.

AI agents can use every staff capability over MCP, under the authority of the person they act for. You decide how far each agent can go, from drafting to signing, and the audit trail always shows who did what. Under the regulated preset, only named people approve and sign, as the draft EU GMP Annex 22 expects for generative AI.

01

Never above the person

An agent acts for one person and gets only what that person may do and your agent policy allows.

02

People confirm what matters

High-risk actions wait for the person to confirm them out of band, with a fresh sign-in. The agent cannot approve its own request.

03

Prompt injection defense in the server

When an agent reads untrusted content, the server lowers its trust level and restricts what it can write. The model does not have to behave for this to work.

04

Every call recorded

Each tool call is recorded in the audit trail with the person and the agent session behind it.

agent session S-7f2 · acting for a person · preset: regulatedillustrative
  1. agentcallcreate_record type=deviation "Temperature excursion, cold room 2"
  2. policyallowwithin the person's own permissions · trust: high
  3. ledgersealentry 48,213 · shard 17 · sha256 9f3a…c21e
  4. agentcalltransition deviation/DV-0193 event=approve
  5. policydenyregulated preset: approval requires a person's signature
  6. agentcallrequest_confirmation command sha256 4be1…07aa
  7. personsignQA lead confirmed · fresh sign-in 42 s ago · meaning "approved"
  8. ledgersealentry 48,214 · signature bound to version digest
  9. ledgerproofcheckpoint signed · RFC 3161 timestamp · verifiable offline

How agents work in Tabellix

02 Extreme control

You set the rules. All of them.

Tabellix is configured at four levels, from the deployment down to individual agent permissions. Changes below the deployment ship as configuration releases that are reviewed as a diff, approved by people other than the author, and recorded on every audit entry.

L1
DeploymentSet at installation and recorded at every startup
Access logging modeSignature re-authentication windowStorage lock modeMinimum retentionEarly destructionBreak-glassAgent presetIdentity providersBranding
L2
Object typePer document, record, or case type
Lifecycle and guardsFields and capture rulesRetention policiesPeriodic reviewAccepted media typesEncrypted contentCase access attributes
L3
ReleaseApproved by people other than the author, applied in seconds
Roles, permits, and denialsAgent allow, confirm, or denySignature meaningsWorkflowsTriggersRedaction codes
L4
Agent presetA regulated baseline, or each permission set individually
Signing, by meaningDestruction approvalSharingHold releaseRelease activationBreak-glass
01

Policy as code

Every access decision is evaluated by policy, denied by default, and recorded with its reasons. An explicit denial always overrides a grant.

02

Existence stays hidden

Users who may not see a record cannot tell that it exists. Search results, lists, and counts treat hidden and missing records the same way.

03

Authorization in the query

Access rules are applied inside the database query, so results, counts, and pages contain only what the user may read.

04

Signatures with meaning

Each electronic signature records its meaning, is bound to the exact version signed, and requires a fresh sign-in.

05

Break-glass, reviewed

Emergency access is time-limited, requires a justification, triggers an alert, and is always reviewed afterward.

06

Scoped external access

Inspectors, auditors, and counterparties get scoped, expiring, watermarked access without a user seat, with a report of everything they accessed.

Security and access control

03 Auditability

Don't trust us. Verify.

Every change is written to a hash-chained ledger in the same transaction. Checkpoints are signed with a key the database cannot use and timestamped by an independent authority. Auditors can verify records, disclosures, and evidence offline, on their own laptop, without access to our systems.

#48210createdprev 00a1…
#48211reviewedprev 9f3a…
#48212signedprev 4be1…
#48213effectiveprev c21e…
#48214disclosedprev 7d0f…
checkpointsignedRFC 3161
✓ verified offline
0failure modes specified, each with defined behavior
0operations available to agents, each governed by policy
0product controls tied to tests and runtime checks
0installation checks that qualify each deployment

Catalog figures from the current release line. They count what is specified, not test results.

Controls map to
  • 21 CFR Part 11
  • EU GMP Annex 11
  • GDPR
  • EU AI Act
  • ISO 13485
  • SOC 2
  • ISO/IEC 27001
  • HIPAA

Product controls map to these regulations and standards and provide evidence for your own validation and audits.

Agent oversight built for
  • EU GMP Annex 22 (draft)

The draft EU GMP Annex 22 keeps generative AI out of critical GMP applications and expects qualified people to stay responsible for AI outputs. Under the regulated agent preset, only named people approve and sign.

Trust and verifiability

04 Privacy and retention

Keep what the law requires. Forget what it allows.

Regulations require you to keep records for years, and privacy laws such as the GDPR require you to erase personal data. Tabellix supports both without breaking the audit trail.

[ Privacy ]
  • A key per person. Personal data about each person your records concern is encrypted with its own key. Erasure destroys the key, and the audit trail still verifies.
  • Pseudonymous attribution. The audit trail names the people who act by stable IDs. Display names are resolved when the history is read, under their own permission.
  • A key per file. Every stored file is encrypted with its own key, and its integrity is checked on every read.
  • Access is logged. Views and downloads are logged before any content is served, at the level you configure. Reads of the personal data held about a person are logged at every level.
[ Retention ]
  • Longest period wins. Retention is set per record type, including permanent retention, above a minimum the deployment enforces. When rules overlap, the longest period applies.
  • Holds stop the clock. A legal hold blocks disposition and destruction, regardless of the retention schedule.
  • People approve disposal. The system proposes disposition, and records administrators approve it.
  • Lawful early destruction. Where the deployment allows it, records can be crypto-shredded early under a legal order with two distinct approvers, never under an active hold, and the audit trail still verifies.

Privacy and retention

05 Industries

Built for the ones who get audited.

One core serves every industry. Each vertical is configured with its own record types, lifecycles, policies, and retention rules, without changing the code.

All industries

06 For software vendors

The record layer inside your product.

Built for vendors of QMS, banking, case management, ticketing, public records, legal, and court software. Embed Tabellix as your record-keeping infrastructure and ship it under your own brand.

01

White label

One setting puts your product name on every customer-visible surface Tabellix provides, from the agent server and confirmation text to evidence titles and webhooks.

02

One API for records

Records, documents, signatures, workflows, retention, and the audit trail are available through one API. You build the user experience, and each vertical is configuration.

03

Every capability through MCP

One contract served over gRPC and REST, and to agents over MCP. Agents reach every staff capability, never with more than the person they act for.

04

Validation evidence, generated

Requirements, traceability, and IQ/OQ protocols are generated from the same traces as the code, with checks that qualify each installation.

05

Any cloud, air-gappable by design

Tabellix runs on Kubernetes with no dependency on a cloud provider. Licenses are verified offline, release verification needs no network, and the software never calls home.

06

Isolation per company

Each company is provisioned with its own database, encryption keys, storage, and audit trail.

OEM and white label

07 Tabellix Cloud early access

The same proof, hosted.

A hosted service for regulated teams that need document and records management with a verifiable audit trail, and for developers whose agents need a system of record. It is an alternative to legacy ECM suites and file-sharing services, with evidence you can verify yourself.

  • EU and US regions
  • A dedicated database, encryption keys, storage, and audit trail for each company
  • A path to your own deployment through a signed, verifiable export
[ FAQ ]

Common questions.

What Tabellix is, how it works with AI agents, and how it supports regulated record-keeping.

01

What is Tabellix?

Tabellix is an agentic system of record for regulated industries. It manages official records, including documents, approvals, electronic signatures, retention requirements, and a tamper-evident audit trail of every change.

02

Who is Tabellix for?

Software vendors that build regulated products, such as QMS, banking, case management, ticketing, public records, legal, and court software. They embed Tabellix and ship it under their own brand. Tabellix Cloud, a hosted service, is in early access for regulated teams and for developers whose agents need a system of record.

03

How do AI agents work with Tabellix?

Agents reach every staff capability over MCP (Model Context Protocol), under the authority of the person they act for and never beyond that person's permissions. High-risk actions wait for a named person to confirm them out of band with a fresh sign-in, and every tool call is recorded in the audit trail.

04

Which regulations and standards does Tabellix map to?

Product controls map to 21 CFR Part 11, EU GMP Annex 11, GDPR, the EU AI Act, ISO 13485, SOC 2, ISO/IEC 27001, and HIPAA. They provide evidence for validation and audits. Tabellix holds no third-party attestation under any framework and claims none; attestation belongs to each deployment and its independent audits.

05

How does Tabellix relate to the draft EU GMP Annex 22 on AI?

Tabellix is built for the human oversight the draft EU GMP Annex 22 expects for generative AI. The draft excludes generative AI and large language models from critical GMP applications and expects qualified people to stay responsible for AI outputs. Under the regulated agent preset, agents draft and named people review, approve, and sign, with every agent action recorded in the audit trail. Tabellix can also hold AI test data and validation evidence under access control, access logging, and retention. Validating an AI model remains the regulated user's responsibility.

06

How can an auditor verify the records?

Every change is written to a hash-chained ledger. Checkpoints are signed with a key the database cannot use and timestamped (RFC 3161) by an independent authority, so an auditor can verify records, disclosures, and evidence offline, without access to the vendor's systems.

07

Are records immutable?

Records are tamper-evident: every change creates a new version and an entry in the hash-chained ledger. Where the storage supports object lock, content can also be written in WORM mode until its retention ends. This is a per-deployment setting.

08

How does Tabellix balance GDPR erasure with record retention?

Personal data about each person your records concern is encrypted with its own key, so erasure destroys the key while the audit trail still verifies. Retention schedules, legal holds, and approved disposition keep what the law requires.

09

Where does Tabellix run?

On Kubernetes, with no dependency on a cloud provider: in a cloud or on premises, and designed for air-gapped networks. Licenses are verified offline, release verification needs no network, and the software never calls home.

10

Can Tabellix be white-labeled?

Yes. One setting puts your product name on every customer-visible surface Tabellix provides, including the agent server, confirmation text, evidence titles, and webhooks. Wire identifiers and signed bytes do not change.

[ Contact ]

Build your product on proof.

Tell us what you build and who audits it. We'll show you how Tabellix fits into your product and how it answers your auditors' most common questions.