Customer Stories
Fedor Zhilkin
Jul 22, 2026
·
Updated on
Jul 22, 2026
Industry: Systems Consulting and Sales Team Development · Application: API integration, automated meeting quality scoring, corporate knowledge base
"I want all of this written down — otherwise how do you process that many meetings? Without mymeet.ai transcripts, my calendar has no life."
— Pavel, B2Boost
About the Company
B2Boost is a systems consulting firm that builds sales departments for its clients. Their differentiator: not to deliver recommendations and walk away, but to fully immerse in the project and drive it to results — from audit through roadmap to a fully operational active sales team. The average engagement runs five months. Target clients are companies with billion-ruble revenues: large retailers and manufacturing enterprises. The team consists almost entirely of former commercial directors; today 8 people actively work through mymeet.ai, including both founders and a partner manager.
619 meetings in April · ~200 hours saved per month · 98% meeting digitization
The Challenge
Pavel at B2Boost manages up to 9 client meetings per day, each for a different client: a billion-ruble retailer one hour, a manufacturing enterprise the next. Between meetings, he must switch completely to a different business, a different project stage, a different client-side team. Holding on to details from two meetings ago is impossible — that context has already been displaced by something more immediate.
Before mymeet.ai, the workflow looked like this: ask permission, start recording, wait 5–10 minutes for processing after the call, download the recording, upload it to a separate transcription service, wait another 20 minutes for the transcript, receive the notes — and only then start working with them. That's dozens of minutes after every meeting; with nine meetings a day, it amounts to a full working shift lost to post-processing.
A second need was growing in parallel. Over a year ago, the team began systematically embedding AI into their work: a significant portion of written content is drafted through GPT and reviewed by a human. For the AI to produce relevant output, it needs context — a lot of client data. That means every meeting needs to be in text form, in one place, with speaker attribution.
"I'd join Zoom, ask permission, start the recording, wait for everything to process, download it, upload it to another transcription service, wait another 20 minutes — and finally get notes I could actually work with."
— Pavel, B2Boost
The Solution
Today at B2Boost, mymeet.ai serves as the platform on which the team has built its own AI infrastructure via API.
Layer 1 — Parallel meeting processing. mymeet.ai is integrated with the calendar and joins every meeting automatically. After each meeting, a concise breakdown arrives via the API in Telegram: what was discussed, which action items belong to B2Boost, which belong to the client. In parallel, the system compares with previous meetings and flags which agreements from last time were followed through — and which weren't. Processing happens during the next meeting: the linear chain of "meeting → 20 minutes of transcription" has become a parallel stream.
Layer 2 — Automated meeting quality scoring. Every project manager's meeting receives a score against the firm's proprietary methodology — for example, 86% compliance. The scoring is built on two weighted parameter groups; the scoring version is tracked in a repository and refined over time.
Layer 3 — Team analytics by period. Once a month, the system generates a report for each project manager: total minutes on calls, number of meetings, average duration, distribution across weekdays. Strengths and weaknesses of each manager are tracked over time — functioning as a signal system for team development.
Additionally, the team uses mymeet.ai to surface methodological violations from transcripts and identify new client needs not yet reflected in the current project roadmap. The entire infrastructure was built by one person using Codex — no dedicated IT team required.
"In my case, mymeet.ai plays the role of an assistant. I literally call it my secretary, 'Mitya.'"
— Pavel, B2Boost
The Results
619 meetings processed in April 2026 by a team of 8 — average meeting length 49.6 minutes, totaling over 31,000 minutes of conversation across the team
~200 hours per month returned from post-processing — equivalent to a full working month for seven people
98% of all meetings digitized — every client meeting and internal session goes into the team archive and shared project knowledge base
The key qualitative shift: B2Boost stopped "logging meetings" and started building a corporate AI asset. Each client in the system is no longer a single document — it's a complete conversation archive that can be queried: "find methodology violations from April," "pull the project dynamics for the quarter." This allows a single manager to simultaneously run multiple engagements with billion-revenue companies.
A side effect has emerged: clients see the bot in meetings, ask "what's that?" — and sign up themselves. mymeet.ai has moved from being a tool to becoming part of B2Boost's partner offering.
"This story spreads itself. Clients see the mymeet.ai bot in meetings — and the reaction is always the same: 'I want one of those too.'"
— Pavel, B2Boost
B2Boost uses mymeet.ai via API to automatically process 600+ meetings per month, score project manager performance, and build a corporate client knowledge base. B2Boost is a mymeet.ai partner.
Fedor Zhilkin
Jul 22, 2026
