Skip to content
Back to work

Automated AI platform

Daily AI Digest Ecosystem

Live automated LLM publishing system with RAG search and AgentOps monitoring.

An automated Live AI intelligence system that generates, publishes, retrieves, and monitors AI/tech updates across an automated LLM workflow.

Daily AI Digest AgentOps Dashboard
Digest items
480+
AI runs tracked
270+
Source-linked RAG
Yes
Errors captured
21
Avg run duration
2m05s
01 /

Problem / Solution

Problem

AI news is abundant, but turning it into a reliable knowledge system is difficult. Information arrives from multiple sources, historical updates become hard to retrieve, and there is rarely a clear operational workflow connecting content generation, publishing, retrieval, and monitoring.

Solution

Built an automated AI publishing ecosystem that turns scheduled LLM runs into structured digests, promotes only successful outputs to production, enables source-linked retrieval over historical content, and surfaces failures through a separate AgentOps dashboard.

02 /

Architecture

Publishing path

  1. 01

    Cron Jobs

  2. 02

    LLM Generation

  3. 03

    Log Parsing

  4. 04

    Data Update

  5. 05

    GitHub Commit

  6. 06

    Vercel Deploy

  7. 07

    Digest Update

Chatbot path

  1. 01

    User Query

  2. 02

    Redis Rate Limiting

  3. 03

    Query Validation

  4. 04

    Embedding / Retrieval

  5. 05

    Top-K Similarity Search

  6. 06

    Context Assembly

  7. 07

    LLM Response

  8. 08

    Source Links / Citations

Failure path

  1. F1

    Cron / LLM Failure

  2. F2

    Error Logs

  3. F3

    AgentOps Dashboard

03 /

Engineering Highlights

  • 1024-dimensional vector embeddings over historical digest items.
  • Top-K similarity retrieval with query validation before LLM calls.
  • Source-linked chatbot answers with citation chips back to original articles.
  • Redis-backed rate limiting for public chatbot usage.
  • Success-only publishing path that keeps failed automation runs out of the public digest.
  • AgentOps dashboard for run history, success rate, failures, duration, latest status, and failure streaks.
04 /

Impact

  • Published 480+ structured AI/tech digest items through an automated LLM pipeline.
  • Tracked 270+ successful scheduled AI automation runs.
  • Reduced manual publishing work through GitHub-triggered Vercel deployment automation.
  • Added RAG-based retrieval so historical digest content becomes searchable instead of remaining static archive data.
  • Added Redis-backed rate limiting and query validation to control cost, reduce abuse, and prevent irrelevant LLM calls.