One engineer and a team of agents

No office, no account managers, no discovery calls before anyone has done anything useful. You talk to the person who does the work.

Who

TezBase is run by Ivan Nedomolkov from Almaty, Kazakhstan. The work before this was business systems: accounting and inventory platforms, integrations between shops and marketplaces, and database work on systems that had outgrown what their owners could manage. Mostly for companies across Central Asia and the CIS, and often the jobs other contractors had already walked away from.

That background matters here for one reason: the work was never about software in the abstract. It was about a person in a company doing something the same tedious way every week, and the question of whether that could stop.

What changed

AI agents made a category of that work cheap enough to sell at a price small companies can pay. Reading fifty pages of vendor terms and pulling out what matters used to be a day of somebody's attention. It is now a task that runs while you make coffee, with a person checking the result rather than producing it.

That is the whole thesis. Not that AI is clever, but that it moved the price of careful reading down by an order of magnitude, and that plenty of expensive human hours sit on exactly that kind of work.

How the work is actually done

A pipeline of AI agents does the volume: reading, extracting, cross-checking, drafting. A human reviews before anything reaches you. That split is why one person can take on work that would otherwise need a team, and it is also the honest answer to «who did this» — both, in a specific order.

Two things follow from it, and they are worth stating plainly:

  • Nothing reaches you unreviewed. Agents are fast and occasionally confidently wrong. A number that goes into a summary has been checked against the source line it came from.
  • Where a figure would be a guess, there is no figure. A summary with an honest gap is worth more than one with a plausible invention in it, because you will act on what it says.

Why it costs what it costs

A review is priced in tens of dollars, not thousands, because the expensive part is automated. There is no discovery phase to bill, no account manager to fund, and no minimum engagement. If the job turns out to be small, it stays small.

The flip side is stated on every page: this reads and organises, it does not negotiate, advise on law or tax, or take responsibility for the decision you make afterwards.