Build an e-commerce agent with OpenClaw
Poster spent 100+ hours and 200M+ tokens building an e-commerce workflow agent with OpenClaw; post focuses mostly on the build effort.
Done withOpenClaw
- What they did
- The poster built an e-commerce agent in OpenClaw on a sandboxed VPS, using Telegram rather than Discord, Kimi K2.5 as the main model, and a cheaper model for heartbeats and background jobs. They gave the agent a linked markdown memory system, sub-agent templates, and decision-tree instruction files it only opened when needed, while they manually restarted the gateway, tweaked configs, and brought in stronger models like Opus or Sonnet when troubleshooting.
- How it went
- After 100+ hours and 200M+ tokens, they say they finally got a workflow running reliably enough to use. The main shortcomings were high token burn, background costs, and destructive mistakes like wiping files or even an entire VPS when permissions and boundaries were too loose.
- Worth knowing
- Their stated operating cost was about $10-15 for a full workday of prompting, plus roughly $1-2/day if heartbeats ran on a pricier model.
Try it yourself with OpenClaw
Help me build a small e-commerce workflow agent for my [store platform, e.g. Shopify] store. Start with one job: [e.g. checking inventory and drafting supplier reorder emails, or updating product listings from a spreadsheet]. Finished means one working run on [3 sample products] with a short summary of what it did. Pause and show me anything before it sends a message, changes a listing, or spends money.
Source: Reddit · Jan 1, 2026
Five of these in your inbox every morning
The best things people got an AI agent to do, each with the prompt to try it.
More like this
Manage 18 agents from a Notion workspace
Notion used as the control plane for a fleet of 18 OpenClaw agents.
Automate freelance admin with a morning briefing agent
An OpenClaw agent sends a Telegram morning briefing, drafts weekly client status emails sent after one-tap approval, and chases follow-ups, cutting admin from about 9 hours to 20 minutes.
Triage failed GitHub Actions runs and self-heal flaky tests
A QA engineer's OpenClaw agent reads failed CI logs, classifies failures, posts analysis to GitHub and Slack and opens issues; flakiness fell from 8% to 0.8%, though it once wrongly altered an assertion.
- Qwen 3.6 27B
Order groceries online with a local LLM skill
The author built a grocery skill for a local Qwen 3.6 27B model that knows their fridge contents, past orders and real-time stock, and it ordered products for them online.
Order dinner by voice via Grubhub and Uber Eats
A writer ordered dinner conversationally with Alexa+ through linked Grubhub and Uber Eats accounts, including customizing and changing the order.
- Claude
Plan and order a week of groceries with Claude
The poster used Claude on Mac and iOS to plan and order a week of groceries, with partial success, and will keep using it up to the grocery-list step.