MMOOZZGG.com

AI Due Diligence

AI readiness assessment: what we check in a company

5 min read2 sources

An AI readiness assessment starts with people, not software: department heads know where time and money are lost. In AI Due Diligence, a full review of a company before decisions about AI, we go through six steps, from interviews to vendor contracts. 54% of organizations name high costs as a barrier to AI, so every step ends with a money figure.

On this page · 6 sections

Interviews with managers

We start with the CEO: what the goals for the year are, which numbers matter, and where the CEO personally sees losses. This talk sets the frame for the other meetings.

Then we talk to the head of each department: sales, purchasing, accounting, support, production. The questions are the same: what work the department does every day, how many people it takes, which numbers the department reports on, and where it waits for other departments.

An interview shows what is not in the rules: workarounds, manual checks in spreadsheets, tasks that depend on one person. These places go first onto the list of candidates for AI.

We also ask about AI: who already uses neural networks, for what, and what came of it. We compare the answers with what the management knows.

Processes and losses

After the interviews we describe the processes that run through several departments: from request to payment, from purchase to warehouse, from customer request to resolution.

For each process we calculate three numbers: how many people work on it, how many hours a month it eats, and what an error or a delay costs. This gives an estimate of the losses in money.

A process suits AI if the work repeats, the rules can be written down and the result can be measured. If there are no rules, we first put things in order: AI will not fix a process that does not exist.

The result of the step: a list of processes ranked by losses. It becomes the first part of the AI solutions map.

Data

AI works with company data, so we check where it is, how complete it is and who has access to it.

We look at exports: how many customer cards in the CRM are filled in completely, whether stock in the accounting system matches the warehouse, whether there is a history of customer correspondence and where it is kept.

HSE ISSEK, based on Rosstat data (September 2026): 34% of organizations name a lack of data as a barrier to AI. So we also look for what lies not in systems but in email and in employees’ personal spreadsheets.

We also mark customer personal data: which of it can be given to AI and where to process it under personal data law (152-FZ in Russia). This step is covered in detail on the page about whether your data is ready for AI.

Systems: CRM, accounting, phone system

We list the systems the company works in: CRM, accounting system, phone system, messengers, email, warehouse and industry software. For each one, we note the owner and the person responsible for access.

The main question: can AI connect to the system without replacing it? We check whether the system can exchange data with other programs, which permissions are needed and what must not be touched.

AI is useful where it sees working information and writes the result back: it opens a deal in the CRM, posts a document in the accounting system, puts a call in a queue. Which connections we build is shown on the page about integrations.

The result of the step: a diagram of which systems are linked, where information is duplicated by hand, and where an employee moves it from one program to another.

Costs of people and contractors

We calculate how much the company pays for work that AI can take: salaries with taxes of employees on routine tasks, contractors, an outsourced call center or accounting.

We also collect IT costs: clouds, neural network subscriptions, licenses, support. This step finds subscriptions nobody uses, and work a contractor delivered that nobody checked.

Costs are needed to calculate the effect: we compare the savings of each solution with what the company spends now. Without this number, you cannot say whether AI will pay off.

54% of organizations name high costs as a barrier to AI (HSE ISSEK, September 2026). The calculation shows whether the costs are justified in your case and on which process the investment returns first.

AI solutions and contracts already bought

We collect everything AI-related the company has already bought: subscriptions, chatbots, AI agents, pilots with contractors. For each solution, we ask which process it covers and what its metric is.

We read the contracts with vendors: who owns the system and the data, where personal data is processed, what happens if you cancel the service, and whether you can take the settings and history with you.

91% of Russian companies work with one AI vendor (a survey by the cloud unit of a Russian telecom group, December 2025). They have nothing to compare its price and quality with, so in the review we compare the terms with the market.

If a solution is only about to be bought, we check it before the contract by a separate scheme: what it does without a person, what it connects to and which metric is in the contract. The scheme is described on the page about checking an AI solution and a contractor.

Sources and fact-checking

Checked by the editors on 13 October 2026.

  1. ComNews: only 26% of Russian companies have an AI development strategy, 22 December 2025 (comnews.ru)
  2. CNews citing HSE ISSEK: AI use in organizations, 24 September 2026 (cnews.ru)

Found a mistake? Tell us

Publications

Mike Mozg

Agentic AI use cases at AT&T: nearly 500 processes now run on agents

AT&T no longer uses AI only to help staff work faster: it is rebuilding whole business processes around agents. SiliconANGLE reported on 14 November 2026 that nearly 500 agentic AI use cases at AT&T are already in production. Here is how the company rebuilds them, where a human still checks the agent, and what other large companies can take from it.

News4 min
Contact us