AI in business: practical uses on top of your ERP

Short answer: AI in business pays off most when it runs on clean data inside an ERP: demand forecasting, automatic invoice reading (OCR), lead scoring, chatbots and plain-language questions about reports. Odoo 19 ships AI agents, AI fields and AI-powered search out of the box. But if your data is a mess, AI projects don't deliver, so processes and data come first.

Almost every manager has heard about artificial intelligence by now: ChatGPT writes emails, draws pictures, answers questions. For a business owner, though, the real question is different: where exactly does AI in business save money or bring in revenue? The answer is often surprising. AI isn't useful on its own; it pays off when it works with the company's sales, inventory, accounting and customer data, and that data lives in the ERP system. This article looks at practical AI use cases for companies in Uzbekistan, what Odoo already offers, and where to start.

What does AI actually do in a business?

AI in business is a set of software tools that learn from company data to forecast, read documents, write text and suggest routine decisions.

In practice, AI does four kinds of work:

Predicts: which item runs out when, which customer comes back, which deal closes.
Reads and understands: a scanned invoice, a contract, a customer email.
Writes and summarizes: a quotation, a reply to a customer, a summary of a long thread.
Converses: a chatbot on the website or Telegram, an internal assistant for staff.

All of this needs context: products, prices, stock levels, customer history. If that data is scattered across spreadsheets, notebooks and people's memories, AI has nothing to stand on.

How widely are companies using AI?

In McKinsey's survey published in August 2026, nearly nine in ten respondents reported regular use of AI in at least one business function. Yet only 44% said AI is scaling across their enterprise, and 37% attributed at least some EBIT impact to AI (McKinsey, The state of AI in 2026).

In other words, starting with AI is easy; making money from it is hard. The same research found that nearly three-quarters of AI high performers fundamentally redesigned workflows because of AI, compared with one-quarter of everyone else (McKinsey). The takeaway: bolting AI onto an old, messy process isn't enough.

In Uzbekistan, AI is a national priority. The strategy approved by Presidential Decree PQ-358 of 14 October 2024 sets a goal of reaching $1.5 billion in AI-based software and services by 2030 (lex.uz, PQ-358).

Why does AI need clean data first?

AI-ready data means records kept in one system, in one format, without duplicates and entered on time.

In 2025, Gartner found that 63% of organizations either lack, or are unsure whether they have, the right data management practices for AI. Gartner predicts that through 2026, organizations will abandon 60% of AI projects unsupported by AI-ready data (Gartner, 2025).

A pattern we often see in projects: the CEO asks for "AI forecasting", yet one product sits in the database under three names and sales are split between the till and a spreadsheet. Any forecast built on that will be wrong. The right order is ERP and a single database first, then AI. More in our article on moving from Excel to ERP.

Which AI use cases on top of ERP actually work?

Demand forecasting and purchasing

Using sales history, seasonality and stock levels, AI suggests what to order and how much. According to McKinsey, AI-driven forecasting in supply chain management can reduce errors by 20–50% (McKinsey, 2022). How inventory data gets captured: Odoo Inventory.

Automatic invoice reading (OCR)

A supplier's PDF or scanned bill no longer gets typed in by hand. Odoo's document digitization service uses OCR and AI to read the document and fill in the vendor bill automatically (Odoo documentation). In Uzbekistan, e-invoices arrive through Didox in a structured format, so they don't need OCR at all. More on automating finance: Odoo Accounting.

Lead scoring and CRM

AI estimates how likely each lead is to close, so sales reps spend time on the most promising customers. In Odoo 19 CRM, the "AI insights" feature gives AI-powered estimates of the probability of winning a lead (Odoo 19 release notes). CRM basics: what is a CRM system.

Chatbots and AI agents

In Odoo 19, AI agents learn from company documents, plug into live chat and create leads from conversations. Internal users can ask an agent about records in their own database (Odoo 19 release notes).

Plain-language search and writing

"Ask AI" search turns a natural-language question into an Odoo search; AI fields and server actions fill in records automatically; AI drafts emails and summaries (Odoo 19 release notes).

What changes after automation: in numbers

Forecast accuracy: AI-driven forecasting cuts supply chain forecasting errors by 20–50% (McKinsey, 2022).
Stockouts: better forecasting can reduce lost sales and product unavailability by up to 65% (McKinsey, 2022).
Warehousing and admin: the same source says warehousing costs can fall by 5–10% and administration costs by 25–40% (McKinsey, 2022).
Invoice entry: instead of manual typing, one document is read via OCR for one credit, and structured XML files need no credits (Odoo documentation).
Bottom-line impact: 37% of respondents attribute some EBIT impact to AI, and nearly three-quarters of high performers redesigned their workflows (McKinsey, 2026).
Risk: 60% of AI projects without AI-ready data are expected to be abandoned (Gartner, 2025).

These figures come from international research; results in any one company depend on its data quality and processes.

Where should a company in Uzbekistan start with AI?

Move processes into one ERP system: sales, inventory, purchasing, accounting.
Clean the data: product records, customer base, units of measure.
Build up at least a few months of clean history, since forecasts rely on it.
Pick one concrete problem: stockouts, manual document entry or lost leads.
Start with Odoo's built-in AI features and consider custom models later.
Measure the outcome: error counts, entry time, share of deals won.

These steps are a natural continuation of digitalization.

How does Deep Vision approach this?

Deep Vision is an official Odoo Gold Partner based in Tashkent. Its Odoo partner page lists 7 Odoo-certified experts (v17–v19), 24 client references and an 89% customer retention rate (Odoo partner page). The team produced Odoo's Uzbek translation: Deep Vision is credited in 360 of the 376 Uzbek translation files in the Odoo 19 source, so AI features also work in an Uzbek-language interface.

In the experience of the team led by Deep Vision founder Amon Olimov, AI projects are built like this:

ERP first: data is brought into one system and cleaned.
Then one concrete task, such as purchase suggestions or automatic document reading.
Built-in tools: Odoo 19 AI agents, AI fields and integrations are configured.
Local integrations: we have ready-made modules for Didox (e-invoices) and Asl Belgisi (digital product marking).
Training: Deep Vision has run 4 hackathons, 3 seminars and an online course, and Odoo is taught with us at 3 universities to more than 300 students a year.

Standard Sales and Inventory are often live in 6–8 weeks; a full rollout takes 2–6 months. Service: Odoo implementation. To discuss AI for your company, fill in the brief.

Frequently asked questions

Does Odoo have AI?

Yes. Odoo 19 includes AI agents, AI fields, AI search, AI writing, voice transcription and "AI insights" in CRM. Document digitization (OCR) is a separate credit-based service.

Does a small business need AI?

It needs an ERP first. Once data lives in one system, built-in AI features help even a small company cut manual work.

Will AI replace employees?

It mostly takes over routine work: document entry, writing emails, finding reports. Decisions and customer relationships stay with people.

Is our data safe?

Odoo lets you choose the AI provider, for example your own Gemini account. Which data may be sent to an external service is agreed at the start of the project.

How long does an AI project take?

If the ERP is already running and the data is clean, switching on built-in features is quick. Without an ERP, that comes first.

Sources

Ready to put AI to work on top of your ERP? Fill in the brief or call +998 77 093 00 07.

Back to blog