We build products, and put AI where it does the work.
We start by looking at how your business runs: the product you sell, the jobs your people do by hand, the product you want to exist. Where a model would do a job better than a person or plain code, we build it in. Where the product doesn't exist yet, we build it, for web, iOS, Android and macOS. Every step has a fixed price.
What we build
A product, built
Web, iOS, Android and macOS, with the backend behind it and a model only where it beats plain code. From a two-week product sprint to a first version in the stores: fixed price per scope, a demo every week, tests from week one, your own repository from day one. Our own apps each went from idea to the App Store in under a month.
Build a productAI in your processes
For owners and COOs. The job one person does by hand every day, handled by a model that reads, sorts and drafts, with that person approving. Built with them, running on your own systems, at a fixed price per tool.
AI in your processesAI in your product
For software companies. One AI feature your customers use, shipped with your engineers under data rules you can put in your terms, and the delivery changes your team needs to keep shipping. Measured in your own Jira and Git.
AI in your productSelling the company? We measure your engineering cost, delivery and key-person risk before the buyer's due diligence does: Before a sale
What you can check yourself
Our own apps, live on the App Store and earning. We built each of them together in under a month, and the list keeps growing. Two of them put a model next to personal data without leaking it.
Anonimo
Blurs every face in a video except the ones you tap. The face model runs on the phone, so the video never leaves it.
Anonimo on the App StoreExpenny
Family budget: say it, photo the receipt, or type it. Receipts and voice notes reach the cloud model with personal details swapped for placeholders.
Expenny on the App StoreProspekt
See what any lens frames on your camera, live. The one without a model: camera work on the phone, and a site in 15 languages.
Prospekt on the App StoreWhat a diagnostic report looks like
Two sample reports. The companies and numbers are made up; the format is the one yours will take.
How we measure lead time · What we access, and when it's gone
How it goes
One or two weeks
We look first. For a product that doesn't exist yet, a two-week product sprint, €8,000: a clickable prototype and a technical plan, and the €8,000 comes off the price of the build. For a product you already sell, a two-week diagnostic, €6,500: where a model fits in the product and the workflow, and what data it would touch. For a job done by hand, a week with the person who does it, €2,000, which comes off the price of the tool; when several people do such jobs, a two-week look at the whole office, €6,500. Each ends with a fixed quote for the next step, and you keep the plan either way.
Four to twelve weeks
The build. A first version of a product takes ten to twelve weeks, €48–60k, paid in four milestones, each after you accept its demo. A pilot that ships one AI feature with one of your teams takes six to eight weeks, €18–30k. A tool takes four to six weeks, from €5,000. Tools and pilots are paid half when we start, half once they have run for two weeks with your people.
Then 60 days
Handover, and 60 days of cover. A runbook your IT team can follow, or playbooks your engineers re-run.
Both of us work on every project
There is nobody else at Nakolini, so nobody else touches yours.
Dmytro Hryb
Engineer · delivery and teamsHas led engineering teams through the changes he now sells: reorganised them into small squads releasing every week or two, brought engineers into product decisions and moved manual QA into automated tests. At Nakolini he leads the diagnostic, the delivery changes and the release rhythm.
Vitalii Kikot
Engineer · AI features and automationBuilds AI features end to end, from the model call to the screen. In Anonimo the face model runs on the phone and nothing leaves it; in Expenny receipts and voice notes reach the model with personal data removed first. He works in front-end and iOS, deep in React Native. At Nakolini he builds the tools and the product features.
Terms
Every step has a fixed price, quoted before it starts, and changes are re-quoted before we do them. A tool or a pilot is paid half when we start and half once it has run for two weeks with your people; a product in four milestones, each after you accept its demo; a pre-sale baseline half at the start and half at handover. You can stop after any step and keep everything. What we write for you is yours, and where a tool uses our shared library you get a permanent licence to it at no charge.
Your data stays yours. Before a cloud model reads an email or an invoice, names and account numbers are swapped for placeholders on your own server, a person on your side approves what it drafts, and the model runs only where you approve in writing, with the endpoint named in your data-processing agreement. What we access, and when it's gone
Our first three clients pay 25% less, in return for one reference call with a prospect, under NDA. Nothing about you is published unless you approve the wording.
Write to us.
Three lines is enough: the job or the feature, what you run it on, when you need it. We both read it, one of us replies within one working day, and if we're not the right people we'll say so on the first call.