One shareable readout that answers that question every morning. It scans the open web and the marketplaces where the category actually moves, scores what matters, translates it into leadership language, and pushes a summary to WhatsApp. This page is the build plan: what is easy, what is high impact, what we ship first, and how fast.
The Daily Intelligence Agent (global brand moves) and the Culture Radar (India youth signal to commerce) are the same pipeline with two output lenses. Build the pipeline once.
Not all sources cost the same to ingest. Here is the honest difficulty of each, from plug-and-play to the ones we phase in later. Tap a tier.
Structured or near-structured sources that need pulling, not fighting. This is where the system earns its keep on day one.
Slightly harder, but this is where the apparel and sportswear signal Agilitas actually cares about lives. Doable with maintained tooling and residential proxies.
Genuinely useful and genuinely harder, so not first. Brittle, against terms of service, and a maintenance cost, which is why we build these in a later phase once the core is proven.
Every candidate source plotted by how much it moves the needle against how hard it is to wire up. The top-left is the fruit worth picking first.
Ship the media and Google Trends radar on day one, then add Flipkart for breadth and Myntra for the fashion signal in week one, on scrapers we build in-house. That single combination already answers the daily question for Lotto, one8 and SportsYard, and can be built quickly and shipped fast. Instagram and TikTok come in a later phase, once this proves out.
Each step is a usable thing on its own. We never build for three weeks and then reveal. Tap through the order.
Pull the fashion, sneaker and retail feeds plus Google Trends into one place, deduplicate, and you already have a real Daily Brand Radar.
Add Flipkart and Myntra so the radar sees not just what brands say, but what is actually selling and trending in the catalog.
An LLM clusters the raw signal, removes duplicates, scores each item and rewrites it the way a CEO reads.
The summary lands in leadership's WhatsApp every morning, with a link to the full view.
The shareable product: Today, Trends, Competitors, Creators, Brand Actions and Archive, with the heat map and per-brand views.
The quick wins roll up into three phases, each with a clear finish line and a thing you can put in front of the room.
A lightly manual but completely real daily radar, fast. The goal is leadership reading actual output within two weeks, not a slide about output.
Take the human out of the daily loop. Automate end to end, formalise scoring, and split the view per brand.
The shareable command center the documents describe, with the full section set and a weekly deck generator.
Almost none of this is built from scratch. Mature open source and ready-made actors carry most of the weight. Where a license or anti-bot tradeoff matters, it is called out.
Maintained Flipkart, Myntra, Ajio and Nykaa scrapers. The single biggest time saver. Data flowing in days, not weeks.
Search, scrape and crawl into clean LLM-ready markdown, with anti-bot handled on the cloud tier. Native n8n and MCP support.
For the few sources that need bespoke handling or cheaper scale. Crawl4AI is permissively licensed, Crawlee brings fingerprint and proxy rotation.
Google Trends interest data and standard RSS feeds. Free, stable, and the highest signal per hour of setup.
Clustering, de-duplication, scoring and the rewrite into leadership language with do now / watch / ignore. The part that turns a feed into a decision.
The daily pipeline as a visual workflow with native nodes for Firecrawl, Apify, the LLM and WhatsApp. n8n is self-hostable.
Airtable to start in days, Supabase (Postgres) as the durable signal archive and the moat over time.
Retool or Framer for a fast internal view, Next.js plus Supabase for the durable, shareable command center.
Meta Cloud API direct (free to host, per-message billing), or Gupshup / WATI / Interakt for easier templates and approvals.
What it costs
Where the risk is, and the answer
Built so there is something real in leadership's hands early, with the full product landing inside a quarter.
This is a draft to argue with, not a finished spec. These are the calls that set the build going the moment they are made.