Agents in the Wild

Reactiv's three-agent AI Scheduler cut merchant app setup time by 80%

Reactiv built three agents that read Shopify store data and draft mobile app updates on a schedule. The company says merchant configuration time fell 80%. Merchants approve changes before they go live.

Organization
Reactiv
Industry
E-commerce software
When
Sep 2026
Evidence
Company reported
Agents and tools
Amazon Bedrock AgentCoreStrands AgentsAmazon RedshiftShopify

Reported outcome

Reactiv reports an 80% cut in merchant configuration time, 33% faster time to production, and scheduled jobs dropping from over 10 minutes to about 5.

  • -80% (company)
    Merchant configuration time
  • 33% faster (company)
    Time to production
  • over 10 min to ~5 min (company)
    Scheduled job run time
  • 17 hours to 3 hours (company)
    Onboarding task effort

Caveats

Stated by the company that uses the agent (blog, earnings call, interview). Not independently audited.

  • All figures are Reactiv's own measurements. The post does not explain how the baselines were measured.
  • The post is co-written with an AWS architect and promotes AgentCore.
  • Nothing goes live without merchant approval, so the agents produce drafts. The time saved is on setup, not on the review.
  • Some items, such as a design system MCP, are plans and not results.

What the agents do

A Supervisor agent reads the merchant's request and decides where it goes. An Analytics agent runs text-to-SQL queries on Redshift to find trends and top products. A Builder agent turns those findings into a new app configuration, using over 50 tools. Merchants give plain-language instructions, such as scheduling a homepage refresh.

What happened

A three-person team shipped the three-agent system in 10 weeks. The earlier single-agent build took 15 weeks. Merchants review each draft before it goes live.

What to copy

  • Split one big agent into a router and two specialists.
  • Let the agent draft; let the customer approve.
  • Run the same agent on a schedule, not only in chat.

Starter prompt

Written by us, not by Reactiv.

Store data summary: {summary}
Merchant request: {request}

Step 1: list the 3 numbers from the data that matter most for this request.
Step 2: propose up to 3 changes to the store's home page. Explain each in one sentence.
Mark every change as a draft for the merchant to approve.

Sources

  1. How Reactiv automates mobile commerce 80% faster with Amazon Bedrock AgentCore — AWS Machine Learning Blog, Sep 22, 2026

Added Oct 26, 2026. Spotted an error? Send a correction.

Get new use cases weekly

One short email. Sources included.

Related use cases