I’ve been experimenting with agent frameworks like OpenClaw recently and wondered how useful that model could be for ecommerce workflows.

A lot of Shopify tasks are repetitive: generating product descriptions, monitoring inventory, sending reports, tagging products, etc. In theory, agent-style systems could automate a lot of that.

The problem I ran into is that most agent frameworks are developer tools, and wiring them directly to a live store API feels risky. You don’t really want an unconstrained agent making arbitrary API calls.

So I built Clawly, which is basically an AI assistant system designed specifically for Shopify stores.

Instead of giving an agent full access, each assistant has scoped permissions and can only use the tools you allow (similar to Shopify API scopes). For example:

- an assistant that only generates product descriptions

- one that monitors orders and sends alerts

- one that produces daily sales summaries

Agents can also connect to external tools like Klaviyo, Notion, email, Google services, etc, so they can run automations across systems.

It’s still early but it’s been useful for automating small store workflows.

Curious what people here think about agent-style systems applied to ecommerce operations.