TLDRocket
Sign in

How Reactiv automates mobile commerce 80% faster with Amazon Bedrock AgentCore

Amazon Web Services Adam Gibicar

Reactiv used Amazon Bedrock AgentCore to automate mobile app updates for Shopify merchants. It cut configuration time by 80% and got the system to production 33% faster.

Based on reporting by Amazon Web Services, Adam Gibicar — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Reactiv builds native iOS and Android shopping apps for Shopify merchants, and that makes timing matter. A homepage that goes stale, or a promo that misses its window, doesn’t just look sloppy. It can leave money on the table, especially when shoppers on mobile convert at 2 to 4 times the rate of web visitors.

The problem was that keeping those apps fresh took a lot of manual work. Merchants had to pick products, shuffle sections, create assets, and push updates on schedule. Most didn’t have the time to do that every week. So Reactiv built an AI Scheduler that lets a merchant say something simple like “Refresh my homepage with best sellers every Monday at 9 AM,” then handles the rest behind the scenes.

That system runs on Amazon Bedrock AgentCore, with three agents working together: a supervisor to route the request, an analytics agent to pull merchant data, and a builder agent to generate the new configuration. Reactiv also uses the Strands Agents SDK, and it got the three-agent setup into production in weeks. According to Reactiv’s internal measurements, merchant configuration time dropped by 80%, and the team reached production 33% faster than with its prior single-agent build.

The architecture is doing a lot of the heavy lifting. AgentCore runtime runs the agents in Firecracker microVMs, so Reactiv doesn’t have to manage containers or scaling policies. AgentCore memory keeps merchant-specific context across sessions, using a session summarizer, a preference learner, and a semantic fact extractor. AgentCore Identity replaced the custom authentication layer Reactiv had built for its Config MCP, and the MCP itself now runs natively on AgentCore as a stateful server that validates every configuration change.

Reactiv also moved its interactive dashboard agent onto the same stack. Before that, the scheduled agent and the chat-based builder were separate systems with separate memory and tools. Now they share the same framework, runtime, memory layer, and UI protocol. The practical upside is simple: what a merchant teaches the dashboard agent can help the scheduler later, and vice versa. The system gets smarter without anyone copying and pasting preferences like it’s 2012.

My take — AI-written commentary, not fact-checked reporting

This is the sensible kind of agent story: less demo, more plumbing. The real win isn’t that an AI can talk about a homepage; it’s that it can safely keep one moving without a human doing digital janitor work every week. Closed systems love flashy assistants, but merchants need boring reliability, not a chatbot with a calendar.

Read more about this at: Amazon Web Services

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.