I built Substation to answer a question a price comparison alone can’t settle: “Where can I pick up these exact glasses tonight, and what will I pay?” A shopping assistant like Muse can delegate that question to Substation. Our agent matches the exact model across merchant catalogs, decides which stores need a follow-up, and saves its recommendation with the sources and timestamps behind it.
In our San Francisco demo, Muse checked Best Buy, Target, and Ray-Ban for a pair of Ray-Ban Meta glasses. On its own, Muse recommended Target. With Substation, it recommended Best Buy on Harrison. The saved receipt showed how a $284 item became a $312.76 estimated total after tax and a recycling fee, and why the shopper should wait for a pickup notice before leaving. Muse’s final reply took 4 minutes 46 seconds on its own and 1 minute 17 seconds with Substation. The assisted run reused retailer checks from the same chat.
Airbyte synced Target’s earlier offer so the agent could distinguish yesterday’s price from today’s. Cognee’s knowledge graph guided the Best Buy pickup check. SingleStore saved the decision and audit trail so anyone could reopen the receipt. Amplitude recorded 18 events and, in a separate test, flagged an unsupported pickup claim.