Give MTS a voice and turn My MTS into a platform

In May 2024 I was brought into My MTS to build a transparent process for embedding products into the company's flagship B2C storefront. Five separate businesses — Fintech, LIVE, Telecom, KION, Shop — compete for entry points in a single app, and the teams need to know ahead of time who slots in where over the coming quarters, why, and how it won't break the logic of the next integrations. On top of that, raising the cross-functionality of the unit's own products.

By that point My MTS held tens of millions of MAU and millions of DAU, a product org of around 200 people, and more than 30 ecosystem products inside. By Markswebb's benchmark it ranked among the top ecosystem apps in Russia, alongside Sberbank Online.

The framing task: bring My MTS to the state of a flagship B2C ecosystem platform. The storefront stays central — it's the surface where the company assembles cross-product interaction across its whole ecosystem reach. Inside it, five separate businesses (Fintech, LIVE, Telecom, KION, Shop) embed by the same rules, with no historical priority for telecom as the "landlord." In parallel I had to stand up a design function where there wasn't really one: voice, the family group with geo-search, savings, the unified complaints center. I came in thinking I'd have it figured out in six months; in reality the first months went to mapping the territory and the politics between businesses, not to design.

I was hired for the "team as a product" angle: unit structure, team assessment, internal community, a public voice to the outside. On top of that I handled targeted hiring through my own network — people who had worked with me before or whom I knew from the industry. AI layered onto that foundation over the last two years.

What mattered most in this period was rhythm. We met often as a whole team: design, PMs, engineering. We pitched new ideas, reviewed concepts and approaches, and after releases we looked at the metrics and talked through what worked and why. We built product interest across the whole team, not just the people drawing screens. Discovery for new products ran on open-data analysis and neural nets, without a classic UX panel — faster and cheaper, and we had an annual plan to hit.

Then came the launches. Across 2024–2025 the team shipped eight public steps in the voice and anti-fraud ecosystem. From "Zashchitnik" (anti-spam with a voice bot) and a smart caller ID to AI noise cancellation, which MTS announced as a market first for mobile voice calls. Plus a separate track around "I don't want to — or can't — take a voice call": Call Recording, MTS Secretary, and "Call to Chat," where the Secretary reads typed text aloud to the caller.

I introduced a 1-on-1 template with every designer: getting to know them, walking the competency map, syncs, growth goals. I used the same map for the quarterly team assessment — a growth plan for each person, not a one-off onboarding checkbox.

The internal community rested on design reviews inside teams — weekly, sometimes biweekly when there were no topics or people were swamped. Attendance was optional, but designers showed up on their own: they brought work before prod, picked up advice, borrowed ready solutions from neighbors, and told the design system what was missing. The lead ran the review; for me it was a litmus test of motivation, engagement, and way of thinking — the kind of thing no report shows.

I launched a discovery process: the PM describes a feature with its corner cases, the designer builds a solution on top. We cleaned up the handoff to engineering. Before me, developers pulled mockups straight from drafts; now only from a prepared space. I didn't get budget for additional art directors, so some products — voice before Mark arrived, savings, the family group — I had to art-direct myself, right alongside the leads.

In practice I didn't build a parallel process of my own — I embedded design into the existing product framework. At the QBR we reconciled the five businesses' priorities against the actual plan; cross-cutting reviews surfaced decisions that broke neighbors; cross-team working groups sorted out specific integrations. My job was to seat design leads at each of these tables as equal participants, not as service staff, and to hold the single My MTS flow together on top of other teams' product decisions: unified navigation, aligned entry points, shared payment and notification flows. For cross-product launches we ran joint hackathons — that format came from me.

We moved the My MTS structure to a platform model: Product Teams, Business Teams (contributor businesses), Inner Source, QBR, Platform Team. A release every two weeks. Roughly a third of the backlog went to core-scenario improvements, the rest to new features and integrations. The agreements didn't always hold: some businesses renegotiated along the way, and we reworked part of the integrations a second time.

Onboarding in 30 seconds. The application pulls through Gosuslugi — no forms, no passport scan, no waiting for approval. Open it, agree, start saving.

100 ₽ minimum. One of the lowest thresholds on the savings-account market. Psychologically it's "worth a try," not "you need to accumulate capital."

Bank-grade yield, without a bank. 16.5% annual at launch — on par with top bank savings accounts of that period. The product runs on MTS digital bonds: for the user it's the familiar savings-account mechanic, for the carrier it bypasses classic banking infrastructure.

UI without noise. The home screen shows the large earnings figure, two actions (Withdraw and Top up), and a "how much will I earn" calculator. The calculator became the product's most visited screen: people opened it before even opening an account, playing with numbers and terms. That "proxy page" strongly lifted conversion from view to real top-up.

To keep it from staying a slogan, I led by example: I built the "Convert" plugin for the routine of preparing assets for content managers — designers used to spend 20–80 minutes per iteration, 4–6 times a month (renaming, converting to JPG/AVIF/WEBP, checking file weight, packing into a zip; plus naming errors and passing the buck), and after the plugin, 1–2 minutes. I showed the team how I did it and why — that's what got them going. We set up an internal sharing group: findings, tips, solution walkthroughs.

From there vibe coding spread in a wave. Building on that, one of the designers — in 8 hours, with no coding background — used Cursor with Claude to build a prototype Cinema 4D plugin that generates system 3D icons in MTS brand style in a few clicks. The author framed the goal like this: simplify and speed up icon creation and make it accessible even to people who had never opened Cinema 4D, so that every icon comes out in one consistent style — the same lighting and materials, with no hand-tuning per scene.

The plugin is triggered with M + T: it loads a base scene with SVG icons already in place; the designer picks an icon, a material, a lighting type, and a rotation angle, sets the save path and file name, hits "Render" — and out comes a finished 3D icon in brand style. After that, assets for the home screen and banners were made in-house by the designers themselves, with nothing handed off to an outside agency. From there the team started spotting its own routines to automate.

In parallel I got into the A/B tests on the mts.ru home page: clean graphics versus graphics with captions. Captions gave a multiple lift in engagement, and the team had to be retrained on the new creative rules.