Video Conferencing Platform with an AI Meeting Assistant

Challenge

The client’s team conducted 6–8 customer calls every day. After each meeting, managers manually reviewed recordings and notes to reconstruct what had been discussed, who had committed to what, and which tasks needed to be assigned to the team. This process took 30–40 minutes after every call.

Tasks were often added to Jira with delays, some agreements were forgotten, and others were documented inaccurately. Clients found themselves repeating the same information during follow-up meetings because decisions from previous calls had not been properly recorded.

The client needed a solution that could automatically listen to meetings, capture agreements, and create tasks in Jira.

Solution

We developed Planaro Meet, a browser-based video conferencing platform that requires no additional software installation. At the core of the product is Planaro Assistant, an embedded AI agent that operates both during and after meetings.

Real-Time Transcription

During a meeting, a worker service receives the audio stream and continuously transcribes the conversation. Every statement is tagged with the speaker’s identity and a timestamp.

The agent accumulates the transcript within the session context and remains available for questions throughout the call. Participants can interact with it via the meeting chat or directly by voice, as if it were another participant in the meeting.

Capturing Agreements and Generating Meeting Summaries

Once the meeting ends, the agent generates a structured summary that includes key discussion topics, decisions made, unresolved questions, tasks, responsible stakeholders, and deadlines.

All information is stored in the organizer’s personal workspace alongside the meeting recording. Instead of scattered notes, managers receive a complete and structured document.

Jira Integration

The agent does more than listen and summarize. It extracts specific agreements from the conversation and converts them into Jira tickets, complete with descriptions, priorities, and assignees.

Task descriptions are generated directly from the context of the discussion rather than from generic templates. Managers only need to review the list, add any additional context if necessary, and approve it. Once approved, tasks are automatically assigned to the responsible team members.

Meeting Archive in MAX

All meeting outcomes are automatically published to a corporate channel in MAX, VK’s national enterprise messaging platform.

Full transcripts, summaries, and task lists are stored there and remain accessible indefinitely. If someone needs to find exactly what was discussed with a client eighteen months ago and who committed to what, a channel search can retrieve the relevant fragment in seconds.

There is no need to manage folders full of recordings or send files manually—everything is already available in a centralized archive.

Results

The time required to process a single meeting was reduced from 30–40 minutes to just 2–3 minutes. Managers now only review and approve the task list before it is sent to Jira.

Agreements are documented accurately and on the same day they are made. Clients no longer need to repeat information during subsequent meetings because commitments are consistently tracked and executed.

An additional benefit emerged that had not been anticipated during the planning stage. Thanks to the meeting archive, teams could instantly locate a specific moment in a conversation: who suggested an idea, when it was said, and in what context.

Exact dates, timestamps, and quotations became available within seconds. This eliminated disputes, reduced emotional disagreements, and protected the client’s team in situations where accurate records were critical.