AI agents in a holding company: what works today and what does not
What AI agents do reliably in a holding company today, where they fail, and six rules that make them safe on group data, with evidence from 2025 and 2026.
For holdings and groups of companies
Sahman connects the systems of every company in your group, brings the numbers to one set of rules and shows you each week what changed and why. Samira, an AI colleague, answers questions about the data.
Weekly review
Company B: returns three times the usual level
What Sahman does
Every company’s data is collected in one place each night. Revenue, margin and clients follow one formula, so everyone in the meeting sees the same figure.
Every Monday a Sahman analyst shows what changed, why, and what to do about it. Not a spreadsheet, but a short review with conclusions.
Answers questions about the data in a minute, spots the unusual on her own and drafts reports. People still make the decisions.
Also: forecasts, a data and AI audit, data governance and team training. All five services
Before and after
Group report
Today
Finance staff merge every company’s files by hand, and the report is ready by the middle of the next month.
With Sahman
Numbers for every company update each night, and the weekly review is ready on Monday morning.
Management meeting
Today
Half an hour goes into arguing whose revenue is right, because each company counts it its own way.
With Sahman
Each metric has one formula for the whole group. The meeting discusses decisions, not numbers.
A problem in one company
Today
A rise in returns at one warehouse is noticed a month later, when it already costs money.
With Sahman
Samira flags the spike the same day and shows where it started.
A question from a leader
Today
To learn the margin of one branch, someone asks an analyst and waits a few days.
With Sahman
Ask Samira in plain words and get the figure and its source within a minute.
These are typical situations in groups of companies, not a description of a specific client.
How it works
Accounting, CRM, bookings, bank statements and the Excel files of every company load into one warehouse automatically. Each company’s data stays in its own country.
Together with your finance team we agree how to count revenue, margin and clients, and write each formula down once. The differences between companies disappear.
What was October revenue?
Group total
4.90 m
Dashboards update every night. Once a week the analyst assigned to your group reviews what changed and why, and flags the events that need a decision.
Returns by week
Company B, illustrative data
week 46214 against a norm of 69
usual level
The AI colleague answers questions about the data, spots the unusual on her own and drafts reports. She uses only checked numbers and never sees personal data.
What changed in sales this week? Group sales grew by 4.2%. One event: returns at Company B are three times the usual level.
How we start
By week 8: the first dashboard and the first weekly review
Interviews with leaders, the decisions that matter, an audit of systems and a baseline to measure against.
Metric definitions, a phased roadmap and an architecture with its costs, signed off by you.
A first management dashboard on a narrow slice of data and the first weekly review.
The full foundation company by company, then forecasts and agents, with a monthly roadmap check.
Security and data
Group layer
agreed aggregates only, no personal data
Azerbaijan
Uzbekistan
Kazakhstan
company data stays in the country
Insights
What AI agents do reliably in a holding company today, where they fail, and six rules that make them safe on group data, with evidence from 2025 and 2026.
Kazakhstan’s personal data law in 2026: storage inside the country, transfers abroad, the new notification regime, fines in tenge and the law on AI.
How a group of companies gets one set of numbers for management reporting: a metric dictionary, shared reference lists, nightly checks and a weekly business review.
A pilot takes four weeks, covers one question your leader cares about and has a written measure of success.
How we count visits
Sahman counts visits anonymously, without cookies.
With your consent we also remember your browser and IP address, count clicks and record sessions, to find errors and make the site clearer.
Sahman counts visits anonymously: no cookies and no stored IP address. If you do not want your visits counted, choose “Don’t count me”.
Cookie policy