BYstudio
RU/EN
Back to selected work

PRODUCT / AI / Web products

AI Lead OS

Find buying requests in Telegram and draft replies for the owner to approve.

BYstudio’s role
Product design · Full-stack engineering · AI integration · Operations
Status
Own software project / public demo. Not for sale.
Open the public demo

Explore the project

AI Lead OS is BYstudio’s own software project. It is not currently for sale or offered for implementation. You can explore a public demo with fictional data.

Open the public demo Discuss a similar system
Public demo.Original screen

The challenge

Buying requests in Telegram need to be found, kept with their original text and followed through as conversations. The owner must control outgoing messages.

Solution

We built one workspace for finding and handling requests. It retains the source, explains the assessment and drafts a reply. The owner reviews the text and separately approves sending.

A request becomes a next action.

Follow the journey from a message to the owner’s decision.

Request

Request

Retain the message and its source.

Assessment

Assessment

Assessment evidence stays beside the request.

Decision

Decision

The owner reviews and approves the draft.

How a request
is handled.

CORE WORKFLOW
  1. 01Telegram
  2. 02Sync
  3. 03Signal
  4. 04Qualification
  5. 05Opportunity
  6. 06Approval
  7. 07Conversation
Sources and discovery
Find public sources, monitor them and deduplicate signals.
Qualification
Interpret the request with context and evidence. Unknown does not mean false.
Operator workspace
Pipeline, messages, states and next actions.
Execution
BullMQ queues, Redis and workers for background operations.
Storage
PostgreSQL with workspace-scoped business data.
Action boundaries
Draft review approval. DNC and source restrictions remain mandatory.

The decisions
that shaped the work.

  • Keep evidence beside the conclusion so the operator can inspect the original context. Missing information remains unknown.
  • Run long operations in workers and BullMQ queues, so the interface does not wait for collection or processing.
  • Scope business data to workspaces. Telegram accounts retain their own limits and operational states.
  • Approve a specific message. Enforce do-not-contact decisions and Telegram restrictions at action boundaries.
  • Cover ongoing conversations, incoming replies and follow-ups, rather than focusing only on new signals.
Opportunity workspace · public demo.Original screen

What we delivered

  • Telegram monitoring and source discovery
  • Signal qualification with retained source evidence
  • Opportunity pipeline, inbox and next actions
  • Background jobs, queues and multiple account support
  • Separate drafting, review and send approval

Result

A deployed product with a public demo using fictional data. Monitoring, qualification, the opportunity pipeline and controlled conversation workflows are implemented. Sending requires explicit owner approval.

Technology

Next.jsTypeScriptPostgreSQLRedisBullMQDockerTelegram API

Need a similar system?

The initial task discussion is free. Describe it in your own words. A BYstudio specialist or Yan Bunin will respond.

  1. Read your description.
  2. Clarify the goal and constraints.
  3. Suggest the scope of the first stage.
Message us on Telegram info@bystudio.ru

For example: @username, a phone number or name@company.com

AI Lead OS is not for sale. Use this form to discuss a separate custom project for your workflow.

Add a detail (optional)

Please do not send passwords, tokens, customer databases or bank files.