HADOOPT
TECHNOLOGIES
Schedule a Tech Consultation
Home
Services
Odoo ERP implementation Rollout, custom modules, migration and support. Web application development SaaS products, portals and internal platforms. Mobile application development Offline-first field apps and customer apps. AI development & automation LLM assistants, document processing, forecasting. Odoo version upgrades ↗ Move to a newer release, scoped and priced on its own site.
Products
OrgExp ↗ Budget, CAPEX and OPEX management.
About Contact Schedule a Tech Consultation
Custom AI development & automation

AI that reads your documents and answers from your data.

Not a chatbot bolted to your website. Assistants, document pipelines and forecasting models wired into the systems you already run — with permissions, citations and a human in the loop where it matters.

Assistant
reads your ledger
Which supplier bills are unmatched this month?
Four, totalling ₹2,84,000. Two have no PO, two differ from the PO by more than your 2% tolerance.
from 4 vendor bills · 2 purchase orders
Flag the two over tolerance for review.
Flagged to M. Iyer for approval. No payment has been raised.
Answers cite their records · actions follow your approval chain
Illustrative.
Where it pays

The work AI should be doing, and the work it shouldn't

High-volume reading, typing and judgement on repetitive data is where this earns its cost. Everything else is usually a report and a better workflow.

Worth building
Reading supplier bills, POs and delivery notes into structured records
Answering questions across policies, contracts and product documentation
Forecasting demand and flagging stock-outs before they happen
Triaging and drafting replies to repetitive inbound requests
We'll talk you out of
AI over data that four systems disagree about — fix the ledger first
Anything that must be exactly right every time with no reviewer
A chatbot on the website when the real cost sits in back-office typing
Low-volume tasks where a report and a workflow change are cheaper
What we build

Four things, each with a number attached to it

Every engagement starts by agreeing what it should save — hours, error rate or turnaround — so there is something to measure it against.

01

Document & invoice processing

Scanned bills, POs, delivery notes and forms read into structured records, matched and queued for review.

·Extraction with confidence scores and a human review queue
·Three-way match against PO and receipt, with tolerance rules
·Writes straight into Odoo or your accounting system
4–8 weeks
02

Private assistants & RAG search

Ask your own documentation and records in plain language, with answers that point at the source.

·Retrieval over policies, contracts, SOPs and live ERP data
·Permissions inherited — people see only what they already could
·Every answer cites the records it was drawn from
5–10 weeks
03

Forecasting & decision models

Demand, replenishment and bottleneck prediction trained on your movement history rather than a generic model.

·Stock-out and overstock warnings against live movement
·Reorder points that adapt to season and lead-time drift
·Backtested against last year before anyone relies on it
6–12 weeks
04

Workflow automation & agents

Multi-step work that used to need a person watching an inbox — with approvals kept where they belong.

·Support triage, routing and drafted replies
·Voice and chat actions for field reps — stock, orders, POs
·Every action logged, reversible and inside your approval chain
4–10 weeks
Guardrails

The parts nobody asks about until month three

These are decided before the first model call, not retrofitted after an incident.

Your data stays yours

No training on your content. Where policy requires it, models run in your cloud tenancy or on your own hardware.

Permissions are inherited

The assistant sees exactly what the person asking can see. Retrieval is filtered before the model, not after.

Answers cite records

Every response points at the documents or rows behind it, so anyone can check it in one click rather than trust it.

No silent actions

Anything that writes, pays or approves goes through your existing chain and is logged and reversible.

Model-portable

Claude, Gemini, OpenAI or open-weight — chosen per task and swappable, so a price or policy change is not a rebuild.

Measured, not assumed

Accuracy and hours saved tracked against a baseline taken before launch, reviewed every quarter.

Tested before anyone relies on it

A written evaluation set from your own documents and edge cases, scored before launch and re-run on every model or prompt change.

You can turn it off

The manual path stays intact behind every automation, so a provider outage or a bad week degrades to how you work today rather than stopping it.

Running cost is designed for

Token and infrastructure spend estimated against real volumes during architecture — the cheapest model that clears your accuracy bar is the one we use.

One owner on your side

A named person in your team is trained on the prompts, the eval set and the review queue, so the system does not depend on us being reachable.

The stack

Multi-model by default

We pick per task and keep you portable between providers — nothing here locks you to one vendor's pricing.

Models
Claude APIGemini APIOpenAI APIAzure OpenAILlama
Retrieval
LangChainLlamaIndexpgvectorPineconeQdrantWeaviate
Serving & data
PythonFastAPIPyTorchPostgreSQLDocker
Integration
OdooMCPn8nWhisperREST / webhooks

AI questions we get asked first

Answered the way we'd answer them on the call.

Ask us something else
Do we actually need AI, or is that the sales pitch?

Often you don't. It earns its cost where there is high-volume reading, typing or judgement on repetitive data. If a report and a better workflow solve it, that is cheaper and we will say so.

Will our data be used to train a model?

No. We use providers under no-training terms, and where policy demands it we run open-weight models inside your own cloud tenancy or hardware instead.

What happens when it gets something wrong?

It is designed to. Low-confidence extractions go to a review queue rather than straight through, answers cite their sources so anyone can check, and nothing writes or pays without your approval chain.

Are we locked into one AI provider?

No. The model sits behind our own interface, so Claude, Gemini, OpenAI or an open-weight model can be swapped per task when pricing, policy or capability changes.

What does it cost to run, not just to build?

We estimate monthly token and infrastructure cost during the architecture stage, against your real volumes — and design around it, since the cheapest model that passes your accuracy bar is usually the right one.

Our data is messy. Is it too early?

Possibly, and that is worth knowing before you spend. If four systems disagree about the same number, the ledger comes first — which is a smaller project than an AI programme built on top of the disagreement.

Bring us the task, not the technology.

Tell us which job eats the most hours and we will tell you honestly whether AI is the cheapest way to fix it.

Discuss your AI project