06 ·Agnidoot RAG· Enterprise

Every document your business owns,
turned into intelligent context.

Not a search box over a folder. An advanced retrieval system that reads your documents, pulls out the entities, processes and figures inside them, connects them to the records already in your Odoo — and makes the whole picture available to your ERP and your AI agents at once.

26 file formats OCR for scans and images Vector + keyword retrieval Embedded on your machines

In plain language

What this actually does for your business.

Skip the technology for a moment. Here is the difference it makes on an ordinary Tuesday.

It's the difference between having all the files — and having someone who has read them all.

Every business already owns the answers. They're in contracts, emails, quotations and QC reports. The problem was never storage. It's that finding anything means knowing where to look, and remembering it exists at all.

Purchase · Tuesday morning "Have we had trouble with this supplier before?"
Today You ask around. Someone half-remembers a delivery problem last year. The email is in a colleague's inbox; the QC report is in a folder nobody can name. You place the order anyway.
With us Three complaints, two late deliveries, and the exact clause in their contract about penalties — with the document and the sentence behind each one.
Sales · before quoting "Have we quoted something like this before?"
Today Your estimator spends the morning searching old files, then rebuilds the quote from scratch because it's faster than finding the last one.
With us The three closest past quotations, what was actually charged, and whether the customer accepted — in the time it takes to ask the question.
Finance · a disputed invoice "What did we actually promise this customer?"
Today The contract was signed three years ago by someone who has left. Somebody reads twenty pages hoping the payment terms are in there.
With us The terms, the amendment nobody remembered, and the clause that contradicts the invoice — quoted, with a link to the page it came from.
Anyone · week one "Who is this customer, and what's the history?"
Today Six months before a new person is useful, because the knowledge lives in other people's heads — and walks out when they do.
With us Everything the company knows about that customer, on day one — the orders, the correspondence, the complaints, the terms. Nothing to ask anyone for.

The pattern in all four: the answer already existed inside your business. It just wasn't reachable in the moment someone needed it — and the cost of that is paid in slow decisions, repeated work and mistakes nobody sees coming.

The idea

One thing in your company that knows everything.

Your best employee is the one who's been there fifteen years — who remembers which supplier was late last winter, what that contract actually promised, and why the customer in Pune gets different payment terms. That knowledge lives in their head, in documents nobody can find, and it walks out at six o'clock.

Intelligent context is that memory, built from your own paperwork — and unlike a person, it reads every document, forgets nothing, and cites where each fact came from.

Contracts & termswhat was promised Quotations & POswhat was ordered Specs & drawingswhat was built Complaints & QCwhat went wrong Email & scanseverything else Your Odoo recordsthe system of record Intelligent Context CITED · RESOLVED entities · relationships · processes rules · figures · dates · obligations Inside Odoocontext on the record Your AI agentsrecords + documents at once Search & workspacesscoped to the asker
It knows who

Your customers and vendors

Every party mentioned across every document, resolved to one entity even when four spellings exist — and linked to the partner record in your Odoo.

It knows what

Your processes and rules

How work actually flows, and the commitments buried in the text — "valid 30 days" becomes a number, and your rules stay separate from your counterparties'.

It knows the numbers

Figures, dates and obligations

Amounts, quantities, deadlines and identifiers extracted as structured values — checkable against the ledger rather than buried in a paragraph.

The pipeline

Seven stages between a PDF and an answer.

Most "chat with your documents" tools do two of these: chunk and embed. The five in the middle are why the answer knows what a vendor is.

01 Ingest 26 formats 02 Extract Tika + OCR 03 Classify type + domain 04 Enrich entities · rules · figures 05 Resolve one entity, many spellings 06 Chunk & embed locally 07 Index vector + keyword stages 04 and 05 are what most retrieval systems skip
01–02 · Ingest and extractGetting the text out

Apache Tika with OCR, plus local fallbacks so a single service being down doesn't stop the batch. Scanned PDFs are detected and sent through image extraction with OCR language headers, rather than returning an empty page.

Twenty-six formats, verified from the code:

.pdf.doc.docx.txt.md .rtf.odt.ods.odp.ppt .pptx.xls.xlsx.html.htm .xml.csv.json .png.jpg.jpeg .tiff.tif.bmp .webp.gif

Green ones are images, read by OCR. A corrupt or empty file is skipped and named in the response; the rest of the batch processes normally, and a server restart mid-job doesn't strand it.

03 · ClassifyWhat kind of document is this?

Every document gets a type, a domain and a confidence score the moment it's processed — no manual tagging needed to get started. A document the system couldn't classify shows an amber "needs review" badge rather than a misleading green "processed", and a correction you make by hand survives a workspace-wide re-check untouched.

04 · Enrich — where the context comes fromThe important one

This is the stage that separates intelligent context from a search index. Each document is read for meaning, not just text:

  • Entities — organisations, people, products, locations and the relationships between them.
  • Identifiers — GSTIN, PAN, CIN and IFSC, genuinely validated rather than copied, with internal consistency checked.
  • Processes — how work actually moves through your business, derived across documents rather than from one.
  • Rules and obligations — as structured values, so "quotes are valid for 30 days" becomes a number a workflow can use.
  • Whose rule it is — your policy kept separate from a supplier's terms, with zero misclassifications across three industries in testing.
  • A quote for every fact — each one traces to the document and the exact sentence it came from.
05 · Resolve — four spellings, one companyThe other important one

"Apex Metering Systems Private Limited", "Apex Metering Systems" and "Apex Metering" are one vendor. Without this stage, a search for one finds a third of the truth.

  • Two passes — shared identifiers first, then fuzzy names.
  • Never merged automatically — near-duplicates are offered for review with a similarity score.
  • Merging preserves evidence — the kept entity inherits every mention; the other is marked merged, not deleted.
  • Linked to Odoo — resolved entities connect to the partner and product records already in your ERP.
06–07 · Chunk, embed and indexMaking it findable

Documents are split into passages, embedded by a model running on the AI-ready box in your building, and written to two indexes at once — a keyword index and a vector index, both isolated per tenant.

Retrieval queries both and fuses the results, so a search finds the exact phrase you typed and the passage that means the same thing in different words.

Keyword index

Finds the phrase you typed

Exact terms, part codes, invoice numbers, names — the things a vector search is famously bad at.

FUSED
Vector index

Finds what you meant

Passages about late delivery, when what you typed was "supplier reliability problems".

And when half of it is unavailable, it says so. A degraded search raises an error rather than returning keyword-only results dressed up as a complete answer. "No results" and "I couldn't look properly" mean very different things to somebody making a decision.

Where the context shows up

Built once. Available everywhere.

The point of building context is not to have a second place to search. It's that everything else in the business gets smarter at the same time.

In Odoo

On the record itself

The vendor you're looking at carries what the documents say about them — terms, history, the complaint from March — instead of living in a folder somebody else owns.

In the agent

One question, two sources

Ask about overdue invoices for customers with quality complaints and get one answer — ERP records and document context together, scoped to what you're allowed to see.

In workspaces

Shared with your team

Group documents by project, client or department, and share a workspace with a person, a team or a group — each with its own permission level.

Getting your documents in

Business documents don't live in one place. They're on laptops, in cloud drives, in object storage and on a server somebody set up years ago.

Google Drive, OneDrive and Dropbox over OAuth, with credentials stored encrypted and state signed.
Object storage — Amazon S3, MinIO and Microsoft Azure Blob.
Google Cloud, plus Calendar and Contacts for surrounding context.
Business systems — Shopify and WordPress.
A server folder the platform can reach.
Direct upload — drop a whole folder in the browser, processed in the background.
Non-destructive sync — a file deleted remotely is flagged for review, never deleted on our side.
Re-uploads never silently duplicate — a disagreeing second pass becomes a review, not a double entry.
Two things to plan around

There is no desktop sync app yet. Device registration exists; a proven path for a file moving from a laptop into the corpus does not. Today the routes in are the connectors above and direct upload — so if your documents live on individual machines, the answer is a synced cloud drive rather than an agent on each laptop. Treat desktop ingestion as unbuilt until someone demonstrates it to you.

There's no file-server connector either — SMB and NFS shares aren't reachable today. And Gmail ingestion was removed in August 2026 because it produced fabricated data; we'd rather ship no connector than a lying one.

Private by construction

Understanding your documents without exporting them.

Turning a document into something searchable means turning it into numbers. Where that happens decides whether your contracts leave the building.

01

Isolation that is a boundary, not a filter

One customer's document surfacing in another's search isn't a bug to fix next release, so isolation sits at the storage layer rather than in a query somebody must remember to constrain.

  • A separate database per tenant
  • A separate search index per tenant
  • Search runs against the tenant's own index
  • No unfiltered path exists to forget a filter on
02

Embedded on the box in your building

The embedding model runs locally, so document text never leaves the deployment — and we hold that as a contract term, not a checkbox, because repointing it would change the answer silently.

  • Local embedding model on your hardware
  • Works air-gapped alongside a local language model
  • Document text never leaves the deployment
  • A contractual setting, not an operational one
03

Visibility that follows your roles

An estate is only as safe as the weakest permission on it.

  • Public, internal or private per document
  • Group-level access on individual documents
  • Workspaces shared with a person, team or group
  • ERP answers already scoped by the asker's own Odoo rights
  • Access refusals surface as failures, never as silent gaps

The details

Questions worth asking before you commit.

Does it handle audio and video?Formats

No — and that was a deliberate decision, not an oversight. Media transcription was explicitly taken out of scope and the code removed. The pipeline handles 26 document and image formats, including scans and photographs through OCR.

If meeting recordings or site videos matter to your use case, tell us during the pilot and we'll be straight about whether it's something we'd build rather than implying it already works.

How is this different from uploading files to ChatGPT?Positioning
  • It knows your ERP. A general assistant can read a contract; it can't tell you which open invoices that contract affects.
  • It builds context, not just an index. Entities resolved, processes derived, rules structured — so an answer knows what a vendor is.
  • Permissions travel with the question. Answers are scoped to what the person asking may see.
  • It stays in your building. Per-tenant storage, local embeddings, text that never leaves.
What's verified versus what's builtHonesty

We separate these deliberately:

  • Verified end to end: per-tenant isolation, ingest and retrieval round trip, OCR on a real image-only scanned document, local embeddings, entity resolution with evidence preserved, rule discovery across three industries, and loud failure on both degraded search and access denial.
  • Built, not yet exercised end to end: per-document visibility levels, group permissions, workspace sharing and collections. Real, and we'll configure and prove them on your tenant before you rely on them.
  • Absent: desktop sync, file-server connectors, audio and video.

No penetration test or compliance certification has been performed on any part of the platform.

Can we run it completely offline?Deployment

Yes. Point both the embedding model and the language model at local hardware and no document text or prompt content leaves your network, including air-gapped deployments. This is the configuration regulated customers should ask for by name.

Every plan includes

Try it free for three months. Keep the support for a year.

No obligation, no card, no lock-in. And what comes with your plan isn't a discount — it's what it takes to make an ERP actually land.

Start free →
Weeks

Start in weeks

Not months. Your implementation is underway in your first week.

1 year

Free support

A full year included with every plan — not a paid add-on.

On-site

Training included

We come to you and train your team in the room, on your data.

100 hrs

Customisation

A hundred hours of our engineering, built into every plan.

No card to start  ·  No obligation  ·  Cancel any time in the three months  ·  Your data stays yours

Questions

Questions about your documents becoming answers.

What it gives your business

What does RAG do, without the acronym?
It makes every document your company owns — contracts, SOPs, policies, manuals, past proposals — answerable. Someone asks a question in plain language and gets an answer with the source quoted, instead of a two-hour hunt through folders.
How is this different from uploading files to a chatbot?
Three ways. Your documents stay yours and are not exported. Answers respect who is allowed to see what. And every answer cites the passage it came from, so you can check it rather than trust it.
Will it show someone a contract they shouldn't see?
No. Retrieval is permission-aware: it can only surface passages from documents that person is entitled to open.
Can we trust the answers?
You can verify them, which is better. Every answer carries citations to the source passage, so the burden of proof sits with the document rather than the model.

How it works under the hood

What happens to a PDF between upload and answer?
Seven stages: it is ingested and extracted, classified, enriched with context, entity-resolved so four spellings of one company become one, then chunked, embedded and indexed.
Does anything leave our network?
Not if you don't want it to. Embeddings can be generated locally and the whole pipeline can run offline inside your own infrastructure — which is the default posture on Enterprise.
What kinds of files are supported?
Documents. Audio and video are deliberately not part of the pipeline — we would rather say so than imply coverage we don't have.
How does it connect to the rest of Odoo?
It is built once and available everywhere: the agents, the MCP door and in-Odoo AI all draw on the same knowledge base, so document context sits alongside the live records rather than in a separate silo.

Getting started

Which edition includes RAG?
Enterprise. It is one of the two products — with MCP — that Enterprise adds on top of Standard.
What should we load first?
The documents people interrupt each other about. Contracts, SOPs and pricing policies pay for themselves in the first week; the archive can follow.

Bring us the question nobody can answer today.

The one that spans your ERP and a folder of documents, and currently costs somebody an afternoon. We'll answer it on your data — and show you the sentence each fact came from.

Book a demo

See it running on your own documents.

Thirty minutes, your paperwork, your questions. We will show you what it does — and, just as importantly, what it refuses to do.

Book a demo →