Private betaStatus
RedisSession store
01

Context

Anonymous matching products need to balance fast pairing with privacy, session continuity, and enough structure to prevent empty conversations.

02

Problem

Coordinate anonymous pair sessions reliably while keeping user identity separated from active conversation state.

03

Constraints

Private identities, expiring sessions, concurrent chats, moderation boundaries, and a low-cost deployment model.

04

Architecture

aiogram coordinates Telegram events, Redis stores active sessions, Postgres-compatible persistence stores durable profiles, and an LLM assists with optional conversation prompts.

05

Decisions

Ephemeral and durable state are separated. Matching remains deterministic and inspectable; AI assists conversation but does not determine identity or access.

06

Outcome

A working private prototype established the session and matching architecture before broader product validation.

07

Learnings

In social systems, safety and state transitions are the product architecture—not secondary moderation features.

08

Defining features

  1. 01Anonymous sessions
  2. 02Smart matching
  3. 03Redis state