Project · LegalTech · AI/LLM · B2B

BLS

An AI platform for legal HR consulting: a consultant bot bounded by the Russian Labour Code and BLS case law, document generation, a knowledge base and B2B workspaces. A move from one-to-one “expert — client” support to a scalable digital service.

PMO · PM · BALLM · RAGIn production · private B2B
Client
BLS — a legal services company
Role
PMO · PM · BA
Status
In production · private B2B
About the project

The project turns legal expertise from a one-to-one support model into a digital platform with AI. It automates typical HR scenarios — dismissals, bonuses, remote work — lowers the cost of a first consultation and raises engagement through content and recommendations.

Functionally it is an AI bot bounded by the Russian Labour Code and BLS cases, with document generation, user file uploads and a materials base: videos, articles, templates. A human expert joins on request.

Goals
  • From a P2P model to a digital solution — less manual work, scalable expertise
  • Automating and lowering the cost of typical HR consultations
  • Accessibility: a low entry price through AI and a subscription model
  • Higher engagement through the knowledge base, recommendations and access to an operator
  • A B2B product: workspaces, access rights, analytics, customization
  • A knowledge base and monetization of the company’s expertise
My role
PMO · PM · BA
  • Requirements and dialogue scenarios: the bot’s knowledge boundaries, handling non-typical requests, hand-off to an expert
  • Tasking for the LLM loop and document generation, backlog prioritization
  • Launch from zero to production: planning, risks, acceptance
  • Manual and automated testing of consultation scenarios
Key functionality
  • AI consultant bot bounded by the Labour Code and BLS cases
  • Document generation from the consultation
  • User file uploads into the request context
  • Knowledge base: videos, articles, templates
  • Roles and B2B workspaces, access rights
  • Request analytics
  • Expert on request and an upsell to manual legal help
  • No long-term storage of personal data
Team
8people on average
  • Frontend×2
  • Backend×2
  • QA
  • Business analystme
  • Designer
  • PMme

I also joined manual and automated testing

Artifacts
  • Labour Codethe bot’s knowledge boundary
  • B2Bworkspaces and roles
  • 0personal data kept long-term
Audience
  • HR specialists and managers who need a quick answer on typical personnel situations
  • BLS client companies with their own workspaces and access rights
  • BLS experts who step in on complex requests
Stack and tools
LLMRAG over the knowledge baseDocument generationB2B workspacesRequest analytics
Outcome
  • Launched from zero to production
  • Legal expertise moved from a P2P model to a digital platform
  • A foundation for the subscription model and selling manual legal help

Some details are under NDA: only public facts are shown.

Other projects