Back to selected work

Case study / Jan.2025 / Dev / Client: CoinAcademy

Crypto News Twitter Bot

I built Crypto News Twitter Bot for CoinAcademy to automate crypto news publishing while reducing noise. Before this, many users relied on continuous feeds like WatcherGuru, where signal and noise were mixed and source context was often hard to follow. The setup was semi-solo: another developer provided a news aggregator API, and I built the rest end-to-end — AI filtering, post generation, publishing workflow, a second reply bot, RAG database, and monitoring. The bot publishes curated updates and a companion account replies to comments with source-grounded context, giving the CoinAcademy audience and broader crypto community a cleaner and more useful information stream.

  • Duration2 weeks setup + continuous improvements
  • Team size2 developers
  • RoleAI Engineer & Bot Developer (semi-solo on core bot stack)
  • Tools usedPython · MongoDB · OpenAI · Vector DB · VPS

01 / Brief

Solution

I built Crypto News Twitter Bot for CoinAcademy to automate crypto news publishing while reducing noise. Before this, many users relied on continuous feeds like WatcherGuru, where signal and noise were mixed and source context was often hard to follow. The setup was semi-solo: another developer provided a news aggregator API, and I built the rest end-to-end — AI filtering, post generation, publishing workflow, a second reply bot, RAG database, and monitoring. The bot publishes curated updates and a companion account replies to comments with source-grounded context, giving the CoinAcademy audience and broader crypto community a cleaner and more useful information stream.

02 / Constraints

Problem

  • Finding a reliable way to interact with Twitter without relying on official API access
  • Maintaining filtering reliability to keep high-signal crypto news

03 / System

Architecture / workflow

  • Automated Twitter publishing for curated crypto updates
  • AI filtering to reduce noise and keep high-signal news
  • Second bot replying to comments with source-grounded answers
  • RAG database connected to CoinAcademy knowledge
  • Monitoring and continuous quality tuning
  • ~10 news posts per day

04 / Evidence

Results / impact

  • Reached 8K followers
  • Published around 10 curated crypto news items per day
  • Automated CoinAcademy’s Twitter information workflow with AI-assisted replies

05 / Gallery

Visual references