Accepting engagements for H2 2026 MSK · UTC+3 · 09:00–20:00
All projects
SP–00 · Principal Holding 🟦 co-founder · operating live · pre-seed 2026 selleros01.ru

SellerOS, the same thing,
written in code.

Twenty-five years rebuilding operations by hand, at Lamoda, PEC, Brinex. SellerOS is the same methodology compiled into a Decision Engine: it knows the real profit on every decision and grows it. Profit-aligned: a share only of the uplift it provably created, not GMV.

  • // stage pre-seed 2026 round
  • // surface 3 WB · Ozon · Yandex.Market
  • // model profit-share a share of the uplift only
  • // technology Decision Engine optimized for margin
// Context

On marketplaces, sellers pay for turnover, not profit. Ads, commissions, returns, fines, every point of GMV eats into margin. Agencies and SaaS vendors take a cut of revenue, they win when you grow, even if you're losing money doing it.

In parallel, the seller faces hundreds of decisions a day: ad bid, the floor price for a promo, re-shelving an SKU, restocking a supplier. All of them taken by hand, on instinct, with a lag.

// Challenge

Write a system that takes decisions every 15 minutes, anchored on a single metric, maximum gross profit. Encode the same operator-grade approach we ran at Lamoda and PEC, with one twist: now it's an algorithm, not a person.

// Decision Engine · formula
EV =
margin
P(fail) × C(fail)
decision for maximum profit, net of the cost of error · selleros01.ru
What we did

Six steps from methodology
to a working product.

  1. 01 / Discovery

    Twenty-five years of operator decisions, distilled into an algorithm

    Walked with marketplace founders through 40+ operating cases: which signals they weigh, how they price risk, what separates «cheap to hold the SKU» from «cheap to move it». Captured 146 rules and 18 ADRs, architectural decisions.

  2. 02 / Decision Engine

    Target function, maximize gross_profit, not GMV

    Unlike classic MMP tools, we don't optimize ROAS and we don't chase rank in search. The objective is gross profit on a trailing 30 days, adjusted for holding cost per SKU and return risk.

  3. 03 / Data layer

    Four data contours, no black boxes

    We pull three marketplace APIs + your ERP/1C. Clean, normalize, align by SKU and warehouse. Every number in the dashboard is clickable down to the source row.

  4. 04 / Profit-aligned economics

    We earn only on the uplift we create

    A base for access to the truth layer + a share only of the proven GP uplift, measured via holdout. The same hash-ledger that issues the invoice proves the value line by line. We physically cannot bill for profit we did not create.

  5. 05 / Live product

    Connected a live WB data stream

    A real seller stream through the GP pipeline: 4,195 events, full profit decomposition, hash-integrity on 100% of records. Fulfillment Profit Router in code, choosing fulfillment for maximum profit.

  6. 06 / Pre-seed 2026

    Now, pre-seed round and first paying customers

    Opening access for sellers from 100M ₽ GMV/year. In parallel, a pre-seed round to scale the team and infrastructure for 200+ customers by end of 2026.

Built

What already runs on real data.

// live product · proven in prod seller data through the GP pipeline
4,195
WB events through GP
full profit decomposition
100%
hash-chain records
integrity line by line
1.23M ₽
historical GP
GMV 5.4M through pipeline
EV
decision engine
margin − P(fail) × C(fail)
We don't optimize CTR. We optimize the money that stays with the seller at the end of the month.
Zafarkhon Alikhodjaev IDEAL Consult · author of «Tasty Stories»

Want to look inside, investor pack of 16 documents?

Pitch-deck (EN+RU), 13-section investment memo, finance model, DD FAQ with 50+ Q&A, customer-dev playbook. Sent under NDA after a short call.