Skip to content
Phase 1 · Early access

Documents arrive from everywhere.
Answers come with proof.

Intellodocs is an AI-native document vault. It captures files wherever they land, reads the content rather than the filename, and answers questions about your archive in plain language — pointing at the exact sentence each answer came from.

  • Cloud or on-premise
  • Your data, your region
  • Answers cite their source
WhatsApp Gmail Outlook Scanner

Example — WhatsApp capture

WhatsApp
Awaiting file
Request

The chase starts where it always does — the client group.

How it works

Capture. Understand. Retrieve.

Three movements, in order. A document is taken in wherever it appeared, read for what it actually is, and handed back with the source line attached when somebody asks. Nobody is trained, and nobody downloads, renames or re-files anything.

01 — Capture

It arrives where it always did

Documents are absorbed where they already land — a WhatsApp group, a forwarded email, a scan from the office printer.

02 — Understand

The AI reads the content

Not the filename. Kind of record, parties, dates and amounts are extracted, indexed, and routed to the right client and matter.

03 — Retrieve

Ask, and get a cited answer

A question in plain language returns an answer bound to its source passage — not a list of ten PDFs to open one by one.

01 — Capture

Captured at the source,
whatever the source is.

A document enters the vault from wherever it already appeared. A client drops it into a WhatsApp thread. A vendor emails it as an attachment. Reception scans a stack of paper. Someone drags a folder straight into the DMS. Intellodocs listens on all four, pulls the file in with its origin, sender and timestamp intact — and nobody forwards, renames or re-uploads anything.

Nothing to install, on either side

No portal to teach a client, no login for them to remember. They keep sending documents exactly the way they do today, and your team keeps working in the DMS.

Messy input is expected, not rejected

Skewed, blurry, badly lit phone photographs go through OCR tuned for exactly that. Where a field genuinely cannot be read it is flagged for review rather than guessed.

Provenance travels with the file

Which channel it arrived on, who sent it and when are recorded against the document itself — so months later you can still show exactly where it came from.

Works on the phone they own

Low bandwidth, entry-level Android, patchy connectivity. Capture still works, because the heavy lifting happens on our side rather than theirs.

02 — Understand

It reads the document,
not the filename.

Every arrival is classified and indexed within seconds, then routed to the right client and matter.

Kind of record
Expense record
Named for retrieval
Travel_Ticket_18Mar.pdf
Destination
Expenses › March 2025
Also extracted
Party · dates · amounts

Tags typed by the user: none

Intellodocs proposing a record type, filename and destination folder inside the chat
03 — Retrieve

Ask the archive
a question.

The vault restates what it understood — the topic and the period — before it answers. Intent is visible and correctable, so search never becomes a guessing game.

Understood as — travel documents filed in August 2026
  • What was the total TDS deducted for Acme Corp in Q2?
  • What are the termination terms in the vendor contract?
  • Which client agreements expire this quarter?
Trust

Every answer points at the sentence it came from.

The answer opens the source document and highlights the exact line. No invented facts, no confident guesses — a professional can verify in one glance and put their name on it.

  • Answers are grounded in the firm's own documents — nothing else.
  • If the vault cannot cite it, it says so instead of filling the gap.
  • Every view, share and edit is written to a tamper-proof audit log.

Asked — what are the termination terms in the vendor contract?

Either party may terminate on 90 days' written notice. Termination for material breach takes effect after a 30-day cure period.

Master Service Agreement · clause 11.2 · page 7
Master Service Agreement — CloudX · page 7

11.1 This Agreement shall commence on the Effective Date and continue for an initial term of twenty-four (24) months. 11.2 Either party may terminate this Agreement for convenience upon ninety (90) days' prior written notice to the other party. 11.3 In the event of a material breach, the non-breaching party may terminate following a cure period of thirty (30) days.

Built for

Deepest for three professions.
Not limited to them.

Chartered accountants, lawyers and company secretaries get document types and extracted fields tuned to their work — that is where Phase 1 goes deepest. It is not a limit: the same vault handles whatever your documents turn out to be.

Chartered accountants

Tax filings, audits, GST compliance and client financials

  • Auto-categorise ITR filings, GST returns, balance sheets, audit reports and Form 16s
  • Extract PAN, assessment year, amounts and TDS details without anyone typing them
  • Surface filing deadlines read out of the documents themselves — assessment years, due dates and periods, not a calendar someone has to maintain

Lawyers & law firms

Contracts, court filings, client mandates and legal matters

  • Auto-categorise NDAs, plaints, affidavits, powers of attorney, court orders and consent letters
  • Matter-wise workspaces — organised by client and case, not by whoever named the file
  • Granular access — juniors scoped to their briefs, partners see everything, clients see only theirs

Company secretaries

Board records, statutory registers and regulatory filings

  • Auto-categorise board resolutions, minutes, statutory registers and ROC filings
  • Version history with visual diffs, so how a draft resolution changed is provable
  • Retention rules per entity, with legal hold on anything under scrutiny

And anyone with the same problem.

If you receive documents from people you do not control, and you are accountable for producing them again later, the vault does the same job — it just learns your record types instead.

Compliance & riskInternal auditOperations & projects Procurement & vendorsHR and people teamsInsurance broking Real estate & propertyGrant and donor reportingFamily offices
The product

Built lean, on purpose.

Phase 1 does four things properly — capture, classification, retrieval and control — rather than a dozen things adequately.

The wedge

Capture on the channel it arrived on

Attach a firm WhatsApp number, a mailbox or the office scanner, and every document sent to it is captured, named, classified and filed — without anyone forwarding, downloading or uploading anything. This is where your client documents already arrive.

Automatic categorisation

ITRs, consent letters, NDAs, court orders, expense sheets and balance sheets are recognised on arrival — no manual tagging.

Metadata extraction

Names, dates, amounts, obligations and identifiers are pulled into structured fields you can filter and sort the whole archive by.

Ask the archive

A question in plain language — “which client agreements expire this quarter?” — returns an answer with the source passage attached.

Multi-channel sharing

Send a document back out over WhatsApp, email or a time-limited link — and keep a record of exactly what went where.

Recognised out of the box

It already knows the documents your practice runs on

Income Tax ReturnForm 16 GST ReturnBalance Sheet NDAConsent Letter Court OrderBoard Resolution Expense SheetEngagement Letter + your own types
Inside the vault

Capture happens on a phone.
The work happens at a desk.

Clients send documents from wherever they are. Your team works in a full workspace — numbered engagements, the folder structure a practice already uses, and permissions down to the individual file.

The workspaces list, one per client engagement A workspace showing folders and filed documents Recent documents across the whole firm Managing users and their roles

Every client and matter gets its own numbered workspace, marked internal, external or under audit — so nothing is ever filed against the wrong entity.

Document control

Everything you would demand
of a serious DMS.

Automatic filing is worthless without governance underneath it. These are the foundation the capture layer sits on, not roadmap items.

Check-in / check-out

An exclusive lock while a document is in use, so two people can never quietly overwrite each other. Every checkout records who holds it and since when, and an administrator can force a release.

Version history & restore

Every revision retained with its author and timestamp. Roll any document back to an earlier version in one click — history is never rewritten, only added to.

Visual diff

Compare any two versions side by side and see exactly what changed — which clause moved, which figure was edited — without opening both files and reading them line by line.

Immutable access log

Every open, download, edit, share and deletion written to a tamper-proof record — who, what, when, from where. Exportable the day an auditor, regulator or opposing counsel asks.

Permissions to the file

Inherit from the folder or override on a single document. View, update or delete, per person or group. External parties see only what you scope them to.

Retention & revocable links

Retention rules per client or record type, with legal hold. Share externally by time-limited, password-protected link — revoke at any moment, and see when it was opened.

Security & platform

Privilege isn't a
feature we added.

Confidentiality, retention and chain of custody were architecture decisions, made before the first screen was designed — and the whole thing runs on channels and infrastructure we already operate.

Your infrastructure or ours

Runs where your policy says it runs

Managed cloud in the region you choose, a single-tenant dedicated instance with its own database and encryption keys, or a full on-premise install inside your own data centre — behind your firewall, air-gapped if required. Same codebase, three deployment models, so data residency is a decision you make rather than one we make for you.

Encryption & key custody

TLS in transit and AES-256 at rest as standard. On dedicated and on-premise deployments the keys are held by you, not by us.

No standing access

Our staff cannot browse your vault. Support access is time-boxed, requires your explicit approval, and is written into your own audit log where you can see it.

Single sign-on

SAML and OIDC, with SCIM provisioning and role mapping from the directory you already run. People join and leave through your existing process.

Exit on your terms

Export original files, folder structure, extracted metadata and the complete access log in open formats. No exit fee, no lock-in period, no held hostage data.

Capture
Meta WhatsApp Business API, email interception and direct upload.
Understand
OCR for phone scans, LLM classification and metadata extraction tuned to legal, tax and compliance records.
Retrieve
A semantic index over the firm's own corpus, with retrieval-grounded answers bound to source passages.
Deployment
Managed cloud, single-tenant dedicated instance, or on-premise behind your own firewall. SAML and OIDC single sign-on, SCIM provisioning.
Compliance
Built to support GDPR, India’s DPDP Act and internal data-localisation policy. On dedicated and on-premise deployments the encryption keys are yours.
Platform
Imbibe's ItNet SaaS foundation — role-based access, versioning, immutable audit logs, retention control.
Phase 1

What you get, and
what isn't built yet.

We're a new product asking you to trust us with privileged client files. The least we can do is be exact about where we are.

Working today

  • Capture from WhatsApp, email and direct upload
  • OCR, classification and metadata extraction
  • Plain-language search with visible, correctable intent
  • Versioning, check-in / check-out, audit trail
  • Per-file permissions and revocable sharing
  • Managed cloud and dedicated single-tenant instances

Early access includes white-glove setup — we migrate your existing folders and configure your client and matter structure with you.

Not in Phase 1

  • Native Gmail and Outlook connectors — next on the roadmap
  • Document drafting or templating
  • E-signature
  • Tally or accounting-software sync
  • A public client list — we have no names to drop yet

Phase 1 firms shape what gets built next, and keep their early-access pricing when it ships.

Questions

The things partners actually ask.

Can we run this entirely on our own infrastructure?

Yes. On-premise deployment installs inside your data centre or private cloud, behind your firewall, and can be air-gapped. You hold the encryption keys, control the backups and decide when to upgrade. The AI components can run against a model endpoint you nominate, including one inside your own network.

Can the AI just make something up?

The vault only answers from your own documents, and every answer is bound to the passage it came from — you see the source line before you rely on it. When there isn't a document that supports an answer, it tells you that rather than filling the gap.

What happens to the years of files we already have?

We migrate them during setup. Point us at your existing folder structure, shared drive or server and we ingest it, classify what's there, and map it onto your client and matter structure before you go live. You review the mapping before anything is committed.

How does WhatsApp capture work without a partner's personal phone?

It runs on the Meta WhatsApp Business API against a separate firm number, not anyone's personal account. Documents sent to that number are captured; ordinary conversation is not stored. Clients keep messaging exactly the way they already do.

If we connect a mailbox, do you read all of it?

No. A connected mailbox is watched for attachments only, and you choose the scope — a shared firm address, specific labels or folders, or particular senders. Documents are captured and filed; correspondence is not stored, indexed or used for anything else.

Does it file things without asking us?

No. Intellodocs proposes the record type, filename and destination, and someone confirms with one tap — you can see that in the capture sequence above. Anything it isn't confident about is flagged for review rather than filed silently.

Our clients send terrible phone photos. Will that work?

That's the normal case, not the edge case. Skewed, blurry and badly lit scans go through OCR tuned for exactly this input. Where a field genuinely can't be read, it's flagged rather than guessed.

What does it cost?

Phase 1 pricing is set per firm based on size and volume, and is locked in for early-access firms when we move to general pricing. We'll quote you on the setup call, before you commit to anything.

What if we want to leave?

You export everything — original files, folder structure, extracted metadata and the full audit log — in a standard format. No exit fee and no lock-in period during Phase 1.

Early access

Give your team back
their 1.8 hours.

Knowledge workers lose close to 1.8 hours a day hunting for information that was already sent to them. Documents stop disappearing, answers come with proof, and nobody has to be a librarian. We're onboarding a limited first cohort with white-glove setup, migration of your existing archive, and pricing locked in for good.

100Firms in the Phase 1 cohort. We'll come back to you within two working days.