Amazon API Data Pipelines & AI Design Workflows — Broad Match Show

In this Broad Match Show, Danny McMillan and Adam Heist cover two of the most practical AI frontiers for Amazon sellers right now: getting direct API access to your Seller Central data and building a fully automated design workflow from inspiration through to live assets.

Adam breaks down how he connected Amazon's SP API and Ads API to an AWS database and wired Claude Code directly to it — giving him real-time, queryable access to years of business data across any metric. No developer required. Danny walks through his 8-step system that takes a seller from a TikTok scroll to a finished, conversion-tested design with brand consistency baked in.

Both share hard-won lessons on where AI gets you (the 70–85% mark) and where the human still needs to step in — plus a candid look at what's changing at Seller Sessions Live on May 9th.

Key Topics

• Amazon API data pipeline — SP API + Ads API → AWS database → Claude Code for real-time analysis
• 8-step AI design workflow — Inspiration capture, memory/photo brain, brand system, mood board, asset generation, build, and quality gate
• CLI vs MCP — Why CLIs are becoming the cleaner integration path for tools like Google Workspace
• Seller Sessions Live (May 9th) — New modular format, no sponsors, £5,000 fine system for service providers pitching
• Health check-in — Adam on fitness goals; Danny on resolving a high ferritin (iron overload) diagnosis

Timestamps

• [00:00] Welcome and introductions
• [01:10] Adam: Getting Amazon SP API and Ads API access as an individual brand
• [05:00] Storing API data in AWS and connecting it to Claude Code
• [07:30] Building custom dashboards and software from your own data
• [09:00] How to approach it if you're not technical — think first, screenshot issues, let Claude walk you through
• [12:25] Danny: 8-step AI design workflow overview
• [13:30] Step 1 — Inspiration capture from TikTok, YouTube, social reels
• [14:20] Steps 2–3 — Memory/photo brain + design system (52 world-leading brands baked in)
• [15:30] Steps 4–5 — TLDraw mood board + asset generation (Nano Banana 2, Gemini, Remotion)
• [17:50] Steps 6–7 — Build stage (React, Tailwind, ShadCN, Netlify deploy)
• [18:30] Step 8 — Quality gate (216-feature scoring: UX heuristics, typography, psychology)
• [19:30] Google Stitch + Perplexity demo: full brand system from a product title + screenshot
• [23:12] Adam: the 70–85% rule and how to think about AI-assisted design cycles
• [27:35] Danny: Google Workspace CLI for email — running launches under 3,000 contacts
• [29:26] Health updates — Adam on fitness; Danny on ferritin/iron overload and phlebotomy sessions
• [35:45] Seller Sessions Live May 9th — format, venue (inside a church), evening networking
• [41:49] The £5,000 fine system for service providers pitching at the event
• [43:01] Wrap-up

Key Takeaways

1. You can get Amazon API access as an individual brand — no developer credentials needed. SP API goes back 720 days; Ads API covers 60 days. Approval takes 1–2 days.
2. AWS as a data warehouse for Amazon data — pipe the API into AWS, connect Claude Code to it, and query anything: anomalies, stock-outs, week-on-week comparisons, year-over-year trends.
3. The non-technical workflow is: think → verbalize → screenshot issues → let Claude solve — you don't need to understand the infrastructure, just be clear on what you want to achieve.
4. AI gets you to 70–85% fast — bring in your designer or team at stage 4, not stage 0. Cycle times drop from 6 weeks to 1 week.
5. CLIs beat MCPs for tool integrations where available — less token overhead, fewer config issues, more cohesive experience in Claude Code.
6. Google Workspace CLI can replace Mailchimp/Klaviyo for small lists — Gmail allows up to 3,000 sends per day; viable for product brand launches under that threshold.
7. Seller Sessions Live is now sponsor-free and profitable on ticket revenue alone — the event model is shifting away from conference-style sponsorship dependency.

Notable Quotes

"Getting the actual real-time API data access has been just another level completely."
— Adam Heist

"The original thought is: I need to get API access and I need to connect that to Claude. That's my thinking. And then you literally just verbalize that and use screenshots as you get stuck."
— Adam Heist

"AI gets you to the finish line faster across way more dimensions, so instead of doing 600 things in a year, you're doing 2,000."
— Adam Heist

"We live in a time whereby execution in a way is taken care of by AI. Where we're needed is on the vision — do we build this or don't we build it?"
— Danny McMillan

"Know with AI it's dumb unless you give it a brain."
— Danny McMillan

Resources Mentioned

• Amazon SP API — Business reports, inventory, listings, SQP data; up to 720 days historical
• Amazon Ads API — Ad performance data; 60-day lookback
• AWS (Amazon Web Services) — Cloud database for storing API data; connects to Claude Code via MCP or CLI
• Claude Code — AI coding assistant used to build the data pipeline and dashboards
• Google Stitch — Free UI design tool; used to generate brand systems from a product image + title
• Perplexity — Combined with Stitch to generate full design systems from Amazon listings
• Nano Banana 2 — Image generation tool controlled via Claude; used in Danny's asset generation step
• Gemini — Used with reference images for asset generation
• Remotion — Video generation component in Danny's design workflow
• TLDraw — Collaborative whiteboard/mood board tool; integrated with Claude for live-updating design boards
• React / Tailwind / ShadCN — Front-end stack used in the build step of Danny's workflow
• Netlify — Deployment target for the build step
• 21st Century Dev / ShadCN MCPs — Component library MCPs used in the build stage
• Google Workspace CLI — Cleaner alternative to Gmail MCP for read+write workflows in Claude Code
• Playwright / Fetch MCP — Browser automation tools; Danny built a 4-stage cascade scraper for Amazon

About the Show

The Broad Match Show is a monthly format on Seller Sessions, hosted by Danny McMillan and Adam Heist. It covers the cutting edge of AI tools, Amazon strategy, and brand building — first Tuesday of every month.

Seller Sessions is one of the longest-running Amazon seller podcasts, hosted by Danny McMillan. Known for deep-dives into conversion, data, and the practical application of AI for e-commerce brands.