NewScriber
Agentic news scraper & editorial audio briefing network. Scrapes tech and business news via Firecrawl, curates & scripts dual-host dialogues using Azure OpenAI, and renders high-fidelity multilingual voices via Gemini TTS.
The Context
In a world overloaded with feeds, news consumption has become a chore. We wanted a hands-free, hyper-personalized, multilingual audio briefing delivered before your first morning coffee.
NewScriber automates the entire ingestion-to-ear pipeline, compiling high-quality dual-host podcast episodes dynamically from real-time tech and business news.
What we built
- Go Orchestration & Webhooks. A high-performance Go engine acts as the central server, handling WhatsApp webhooks (Twilio/Meta), PostgreSQL user persistence, and serving a beautiful visualizer dashboard.
- Multi-Agent n8n Pipeline (ReAct + evals). A cron-driven workflow parses uncached articles using Firecrawl, then runs ReAct-style reason-act loops over Azure OpenAI to filter, rank, and summarize the best 7-10 stories and script natural alternating dialogues — with grounding evals gating every script against the scraped source bodies before it reaches synthesis.
- High-Fidelity multi-speaker TTS (Google Gemini). Renders dual-host dialogue through Google's Gemini TTS multi-speaker mode — native voice profiles per language (Alex & Sam for English, Marie & Pierre for French, Sofia & Marco for Italian, Nusrat & Fahim for Bangla) with speech-style modifiers, interruption/turn-taking, and temporal + contextual continuity so the two hosts actually sound like they're in one conversation.
- FFmpeg audio stitching. Concatenates the per-turn 16-bit PCM streams @ 24kHz into seamless long-form episodes via ffmpeg, then transcodes them into WAV, MP3, and OGG stored in Cloudflare R2, generating self-healing iTunes-compliant RSS feeds.
- Visualizer Workspace. A responsive control center dashboard allowing users to trigger drops, play episodes, review transcripts, and see summarized source attributions.
- Automated Podcast Delivery. Delivering high-fidelity multilingual audio drops directly to modern audio spaces like Spotify and Apple Podcasts—all fully automated and regionally scheduled for perfect morning timing and fresh daily source ingestion.
What we learned
Built in a high-intensity sprint for the AI Agent Olympics (Milan AI Week Hackathon 2026), NewScriber offers a zero-app, zero-scroll, zero-noise morning news ritual.
The platform proves that complex multi-agent orchestration, high-fidelity TTS synthesis, and containerized operations can be combined into a seamless consumer-ready product in a single week.
Collaboration & Origin
The core product concept originated from Fauzul, who envisioned a hands-free, hyper-personalized audio news briefing. Asif validated the market need, supported the strategic push for the Milan Hackathon, and actively shaped the MVP scope.
Wasif provided vital API infra support for dev-to-prod agility and shared critical architectural insights from his past work with web scraping and text-to-speech pipelines to speed up delivery.
High-calibre architecture decisions.
Technical Highlights
- ReAct reason-act agent loops with grounding evals gating every script to the scraped source bodies.
- Multi-agent n8n workflows integrating Firecrawl for datascape extraction, Gemini TTS, and ElevenLabs to synthesize brand-specific context and sponsorships.
- Google Gemini TTS multi-speaker synthesis — speech-style modifiers, interruption/turn-taking, temporal + contextual continuity across the dual-host dialogue.
- FFmpeg concatenation of 16-bit PCM @ 24kHz into long-form episodes, transcoded to WAV/MP3/OGG.
- Zero-config Docker deployment for rapid scaling across environments.
Key Decisions & Rationale
Required a high-performance, concurrent engine to handle multiple TTS generation streams and WhatsApp webhooks simultaneously without latency bottlenecks.
Using a visual workflow builder allowed the team to iterate on agent logic (filtering, ranking, scripting) in real-time during the hackathon without recompiling core services.
By the numbers.
Visual & Audio Synthesis Showcase
Experience the alternate voice chemistry and responsive design. Stream pre-synthesized audio drops generated dynamically.
NewScriber Visualizer Workspace Dashboard
If we worked together.
I bring high-leverage product engineering and absolute operational discipline: shipping weekly, pressure-testing user inputs, and automating delivery pipelines.
See it in the wild.
This case study covers how NewScriber was built — these links open what ships today: the live product, repo, demos, and supporting material.
The changelog.
WSL Linker
Open-source Windows GUI + CLI that links Windows Host directories straight to the WSL 2 Linux filesystem via native directory symlinks. Born the week Google's Antigravity IDE 2.0 broke the WSL Remote workflow — built to restore zero-latency, native-Linux-speed cross-platform development on a Windows desktop.