The market told me my price was wrong — in the other direction
Today I went through the AI job listings of five of the largest Russian tech companies — Sber, Yandex, T-Bank, MTS, Avito. Looking at where the market is going. All five are building the same thing: in-house AI-adoption teams per business line — agents in production, answer-quality evaluation, governance. Only Sber publishes salaries, and only for architect roles: from ₽400,000 a month for an AI architect in B2B and up to ₽450,000 for a GenAI architect. The others don't show numbers, but they're hiring the same teams.
The second public number: the partner appliance built on Yandex's stack (Yandex Cloud + Selectel + MetaMentor) puts a typical three-month "AI inside your perimeter" pilot at around 5.8 million ₽. Official estimate — Forbes, end of May.
My two-week Trusted AI sprint costs 450,000 ₽. Fixed. For six months I thought that was expensive for a first client. It turns out to be roughly one month's salary for one AI architect that the country's largest bank is hiring right now. Or roughly one twelfth of that pilot.
The price was never the problem. The problem was that I never said it to anyone: zero proposals sent. A price nobody has heard is neither high nor low.
What changed in a day
I fixed the ladder: trust audit 150,000 ₽ (one week, a passport-style report, credited toward the sprint) → sprint 450,000 → support 150,000 a month with monthly re-acceptance. All fixed prices, no "from".
And I removed the cloud tariffs — 10,000 to 70,000 ₽ a month — from the site. I did the arithmetic honestly: to match what four support clients bring in, the cloud tier needs forty to sixty paying organisations. That's a different company — self-serve funnel, support desk, marketing budget. My product lives on how deep I go into a client's corpus, and depth doesn't scale to a crowd. Few clients, deliberately.
The cloud stayed as a demo surface: to show, on a call, how citation and honest refusal actually behave. Showing is free. Access, deployment, support — not.
What I took from it
Job listings at large companies are free market research that nobody reads as research. They state what the business is ready to buy, in which words it names the thing, and how much it will pay for it internally. I spent three months phrasing the product in my own words. The market had been phrasing it in job descriptions the whole time.
If you're unsure about your price — first check whether you've ever said it out loud.