Skip to demo
SECURE DOCUMENT AIBUILT IN CANADARUNS INSIDE YOUR PERIMETER

Every record.Only for the cleared.

Tabellar turns paper archives and scattered files into a knowledge layer your AI agents can search. It runs on infrastructure you control. Each answer shows only what the person asking is cleared to see.

Book an AI adoption briefing Try the demo
VIEW AS
RECORD · BOX 112 · P. 4SCANNED 2026-09-12 · SHA-256 9F3A…C21
  1. INSPECTION REPORT · SUBSTATION 14
  2. Date of inspection: 14 October 1987
  3. Inspector: R. Okafor, Facilities Division
  4. Transformer T-2: oil seepage at the lower gasket. [1]
4 OF 7 LINES SHOWN3 WITHHELD
QUERYWhat did the 1987 inspection of Substation 14 find, and what was recommended?
ANSWER · FIELD STAFFThe October 1987 inspection found oil seepage on transformer T-2 [1]. What was recommended is above your clearance.
  • [1] Inspection report · Box 112 · p. 4, line 04SCAN
LOGGED 14:02:11 ET · ROLE FIELD STAFF · 3 LINES WITHHELD · AGENT courier-07

From the filing cabinet to an AI agent, and the record never leaves your custody.

PAPER → AGENT · END TO END
01 / THE CUSTODY CHAIN

Six steps.
One unbroken seal.

Others start with data that's already digital, or stop at extraction. Tabellar covers the whole path, from boxes of paper to an agent that answers with sources.

  1. 01CollectBoxes of paper, legacy drives, shared folders and email archives, inventoried with chain-of-custody records from day one.
  2. 02DigitizeScanning, OCR and handwriting recognition, then classification into structured records your teams already understand.
  3. 03SealEncrypted and stored where you decide: air-gapped, on your servers or in Canadian sovereign cloud. Existing permissions carry over.
  4. 04CourierEach agent carries its own identity, traceable to a person, and retrieves only what that person is cleared to see.
  5. 05CiteEvery answer links to the exact page and the original scan. No source, no answer.
  6. 06WitnessTamper-evident logs of who asked what, when and why, exported straight to your SIEM.
02 / THE ONTOLOGY

Not a pile of PDFs.
A model of your organization.

Tabellar maps every record onto the things your organization actually runs: sites, assets, cases, contracts and people. Agents follow those relationships instead of guessing from text fragments.

  1. SITESubstation 14
    HAS ASSET
  2. ASSETTransformer T-2
    INSPECTED IN
  3. INSPECTION14 Oct 1987
    TRIGGERED
  4. WORK ORDERWO-89-0331
    SIGNED BY
  5. PERSONR. Okafor

Entities, not chunks

Each page is linked to the site, asset or case it describes, so related records come back together.

Relationships agents can follow

“Which assets inspected by this team were later replaced?” becomes a path through the ontology, not a keyword search.

Clearance on every node

Permissions apply to entities and links too, so an agent can’t infer what it isn’t allowed to read.

03 / YOUR EDGE

Your archive is your moat. Not someone's training set.

AI labs now pay for private company data, and brokers are lining up to sell yours. We go the other way. Decades of inspections, decisions and case files become an advantage only your organization can use.

NEVER SOLD

Your data is not our product.

We earn from software and services. We never sell or license your records to AI labs, brokers or anyone else.

NEVER SHARED

No pooling across clients.

Each ontology is yours alone. Nothing learned from you reaches anyone else.

NEVER LEAVES

Inside your perimeter.

Records, embeddings and logs stay on the ground you chose.

LAT. · NOUN · MASC.ta·bel·la·ri·us

1. A courier who carried sealed tablets between government offices.
2. One trusted to deliver a record unread and unaltered.

04 / WHY TABELLAR

We took the name. And the standard.

Rome's couriers proved a record was intact by the seal on it. Tabellar does the same for the age of AI agents.

Sealed at the source

Encrypted the moment a page is read.

Unbroken in transit

Never leaves the ground you choose.

Verified on arrival

Cited, cleared, logged.

05 / TRUST LEDGER

No vague promises. Here's exactly where we stand.

DEPLOYMENT
Air-gapped, on-premises or Canadian sovereign cloud. Chosen per workload.
IN BUILD
DATA RESIDENCY
Storage and inference both stay in Canada. Nothing is routed to foreign model endpoints.
IN BUILD
ACCESS CONTROL
Permission-aware retrieval. Each agent has its own identity and is checked on every request.
IN BUILD
AUDIT
Tamper-evident logs of every query and every document touched, ready for your SIEM.
IN BUILD
DATA OWNERSHIP
Your records and your ontology belong to you. We never sell them or share them across clients.
BY DESIGN
COMPLIANCE
Designed around ITSG-33 controls. Protected B / PBMM assessment.
[STATUS]

BY DESIGN · SHIPPED  /  IN BUILD · COMMITTED, NOT YET SHIPPED  /  [STATUS] · NOT YET ASSESSED

06 / WHO IT'S FOR

Organizations whose records can't leave the building.

  • Federal & provincial ministries
  • Government departments & agencies
  • Municipalities
  • Crown corporations
  • Utilities & energy
  • Mining & manufacturing
  • Insurance & financial services
07 / TABELLAR ACADEMY

Your team learns the craft, not just the tool.

AI adoption briefings for leadership, hands-on workshops for records and IT teams, and a readiness review before anything ships.

Explore the Academy →