AI-Powered Buy (aipb)
Visit Live SiteAn AI research agent that finds the best software and AI tools for any job. It fans out across Google, Bing, and Reddit in parallel, resolves contradictions between sources, and writes a cited, ranked comparison article in about 22 seconds.
This project aims to remove the guesswork from buying decisions, turning hours of tab-hoarding and forum-diving into a single, sourced answer.
Why I Built This
New digital products and AI tools launch every day, and at work and in daily life we're constantly searching for the best fit among them. Finding it usually means reading blog roundups, watching YouTube reviews, and digging through Reddit threads, and even after all that, one article rarely gives the full picture. Worse, a lot of "best of" rankings are quietly shaped by affiliate commissions, with the highest-paying product placed first instead of the best one. I built AI-Powered Buy to do that research for you: it reads the blogs, reviews, and Reddit discussions in parallel and hands back a cited, unbiased answer in about 22 seconds, instead of an afternoon.
My Contribution
As the founder and full-stack/AI engineer, I designed and built the platform end-to-end, including:
-
Designed a multi-source research agent
using Python, FastAPI, and LangGraph that runs Google, Bing, and Reddit searches in parallel via BrightData, then synthesizes the findings with OpenAI GPT into one balanced answer.
-
Built the React & TypeScript frontends
covering the public site and the research flow, plus a separate full admin dashboard with revenue, cohort, and funnel analytics, subscription and billing management, affiliate link tracking, and live agent health and research run monitoring.
-
Built the NestJS API layer
handling auth, subscriptions, and orchestration between the frontend and the Python research agent.
-
Integrated billing and deployed the infrastructure
using Dodo Payments for subscriptions and Resend for transactional email. The frontend is deployed on Cloudflare Pages while the API, agent, and Postgres/Redis run on a DigitalOcean droplet.
Technologies Used
-
Frontend: React (Vite) & TypeScript
-
Backend: NestJS
-
AI Agent: Python, FastAPI, LangGraph & OpenAI GPT
-
Data Sources: BrightData (Google & Bing SERP, Reddit)
-
Database: PostgreSQL & Redis
-
Payments: Dodo Payments
-
Security: JWT auth, Passport, bcrypt & Helmet, with NestJS rate limiting
-
SEO: Per-page meta tags, JSON-LD structured data, sitemap & Open Graph cards
-
Date
:
Jan 2025
-
Categories
:
AI Agents & Web Development
-
Client
:
Ashen Thilakarathna (Myself)
- Website :