AI · 2026
Frydai
A private AI "operator" you run from Telegram — it handles competitor research, ad creative, landing pages, and Shopify ops for e-commerce brands.
Problem
Running a small e-commerce brand means wearing every hat at once — media buyer, competitor researcher, creative designer, store manager — usually late at night. Most "AI tools" hand you yet another dashboard to learn and babysit. Frydai's premise is the opposite: the work should show up finished, waiting in your chat.
What it does
Frydai is a private AI operator you talk to from Telegram. It plugs into a brand's own infrastructure and runs jobs end-to-end:
- Competitor intel — scrapes dozens of competitors for angles and offers.
- Ad creatives — generates campaign visuals ready to ship.
- Landing pages — produces conversion-focused pages.
- Shopify ops — manages product listings and order workflows.
Each customer gets their own provisioned bot instance, wired to their Shopify, Meta, TikTok, Pinterest, and Airtable accounts.
Stack & architecture
A Next.js 16 / React 19 dashboard fronts the product; Supabase (Postgres) holds auth and customer state. Membership and checkout run through Whop with an OAuth 2.1 + PKCE flow. Per-customer bots are provisioned as isolated containers via Coolify, each holding that brand's API connections. The surface area is wide — Shopify, Meta, TikTok, Pinterest, Airtable, plus Apify for scraping — so the integration layer and webhook/cron plumbing is where most of the engineering lives. Playwright and Vitest cover the flows.
Status
Pre-launch, founding-access. My role spans the dashboard, the auth and membership lifecycle, the integration layer, and bot provisioning, working with a small distributed team. No live demo — it's customer-infrastructure software.