🎮 The Next Input — Issue #064

The AI That Does Your Company's Shopping

Aaron Bost
September 30, 2025

In partnership with

⚡ The Briefing — 60 sec

OpenAI takes on Google & Amazon with new agentic shopping system. Maybe the new Pulse will go straight to my favorite videogame checkout page…
Anthropic launches Claude Sonnet 4.5, its best AI model for coding. It’s a powerhouse—something that took me a week to piece together was built in 45 minutes. No exaggeration.
AI “actress” Tilly Norwood sparks Hollywood debate. Oh boy. AI really is coming for jobs…

🛠️ The Playbook — AI Shopping Agent for E-Commerce

Mission Deploy an AI-powered shopping assistant that handles discovery, price comparison, and purchase workflows—while respecting budgets and approval policies.
Difficulty Advanced | Build time 3–4 hours (pilot)
ROI Saves procurement/ops teams ≈ 10–20 h/week and reduces overspending by up to 25%.

0) Why This Matters

With OpenAI moving into agentic shopping, the race to automate B2C and B2B purchasing is here. A lean version built inside your org today gives employees a trusted shopping copilot that doesn’t just buy—it buys smart.

1) Architecture

Layer	Tooling	Purpose
Trigger	Slack command `/buy` or browser extension	Start purchase flow
Search & Compare	APIs (Amazon, Google Shopping, Walmart)	Pull price/options
Processor	Claude Sonnet 4.5 / GPT-4o	Parse specs, compare features, rank
Policy Engine	Airtable/Supabase	Store budgets, approved vendors
Executor	Selenium / API calls	Auto-fill carts, generate purchase requests
Review	Slack interactive card	Manager approves/denies
Memory	Supabase	Store vendor history, recurring buys

2) End-to-End Workflow

Trigger
- Employee types /buy Logitech MX Keys in Slack.
Search
- Agent queries Amazon/Google APIs → fetches {product, price, vendor, rating, shipping}.
Compare
- LLM creates ranked table: best price, fastest ship, vendor compliance.
Policy Check
- Cross-check against Rulebook: budgets, approved vendors, spend category.
Decision
- If within policy → auto-generate purchase order draft.
- If exception (e.g., non-approved vendor) → route to manager.
Review
- Slack card → “Buy / Reject / Modify.”
Execution
- On approval, agent fills checkout/cart or sends order to procurement system.
Memory Update
- Logs vendor & price → improves next search.

3) Example Prompts

Comparison Prompt

SYSTEM: You are a procurement analyst. 
INPUT: {product search results JSON} + {policy rules}.
TASK: Produce:
- Ranked top 3 options with vendor, price, rating, ship time.
- Policy compliance: true/false + reasons.
- Recommend “best overall” with justification.
Return JSON only.

Slack Card Output

### Purchase Request: Logitech MX Keys
- Best Price: $89 (Amazon) ✅ Approved Vendor
- Fastest Ship: $92 (Walmart) 2 days
- Compliance: ✅ within $100 budget
**Action:** [Buy Amazon] [Reject] [Modify]

4) Guardrails

Budgets: Max spend per category/employee.
Vendors: Whitelist of approved suppliers.
Risk: Auto-flag “too cheap” deals (<30% market avg).
Human-in-Loop: All exceptions require manager click.

5) Pilot Rollout — 2 Hours

Build /buy Slack command → Make/Zapier → Claude compare.
Fetch Amazon results via API.
Cross-check Airtable “Rulebook” (approved vendors/budgets).
Post first Slack card.
Run for 1 week with office supplies category.

6) Metrics

Avg hours saved per request.
% requests auto-approved (vs exceptions).
$ saved from vendor/price optimization.
Cycle time from request → checkout.

Pro tip: Use this for recurring purchases (office supplies, SaaS renewals, event gear). Once trust builds, extend to high-value buys with stricter checkpoints.

🎯 The Arsenal — Tools & Prompts

Asset	What it does	Link
Amazon Product API	Fetches live product/pricing data.	https://developer.amazonservices.com
Google Shopping API	Multi-vendor product search.	https://developers.google.com/shopping
Supabase	Policy/rule + purchase memory.	https://supabase.com
Prompt · Policy-Aware Compare	Returns ranked purchase list.

Compare these results. Return JSON {vendor, price, compliance, reason, recommended}.

💡 Free Office Hours

Want to build an AI shopping agent for your org?
Book a free 15-minute Office Hours slot—no sales pitch, just workflows solved.

→ Grab a slot: https://calendly.com/aaron-cylentis/the-next-input-office-hours

Become An AI Expert In Just 5 Minutes

If you’re a decision maker at your company, you need to be on the bleeding edge of, well, everything. But before you go signing up for seminars, conferences, lunch ‘n learns, and all that jazz, just know there’s a far better (and simpler) way: Subscribing to The Deep View.

This daily newsletter condenses everything you need to know about the latest and greatest AI developments into a 5-minute read. Squeeze it into your morning coffee break and before you know it, you’ll be an expert too.

Subscribe right here. It’s totally free, wildly informative, and trusted by 600,000+ readers at Google, Meta, Microsoft, and beyond.

🕹️ Game Over

Automate one purchase request tonight—tomorrow your ops team won’t waste hours comparison shopping.
Share your win; you could headline Issue #065.

— Aaron
Automating the boring. Amplifying the brilliant.

Forwarded this? Subscribe here