Skip to content
← Back to Insights

Meta Expands Muse Shopping: Fewer Steps Still Need a Clear Comparison Scope

AI Meta Muse AI Agents Product Design News

TL;DR

Meta adds 10 retail connections at Connect, Spotify announces its Muse integration, and glasses support remains a coming-months plan. Easier purchasing does not establish that an agent compared the whole market.

Meta Expands Muse Shopping: Fewer Steps Still Need a Clear Comparison Scope

Meta expanded Muse’s shopping connections at Connect. That makes me more willing to use it to purchase products I have already chosen. Before asking it to decide which store offers the best deal, however, I would need to see which stores its comparison covers. A complete comparison record could justify delegating more of that decision.Announcement

The keynote was scheduled for 16:00 Pacific on 2026-09-23, equivalent to September 24 at 07:00 in Taipei. This article is dated September 24 in Taipei, with a reporting cutoff of 12:47: it covers that day’s event rather than recycling an older announcement. Meta’s announcement supplies only a date. TechCrunch’s timestamp is 2026-09-23, 18:13 PDT, or September 24 at 09:13 in Taipei; publication time does not establish when individual features became active.Official schedule Independent report

The announcement lists 10 additional retailers, including Walmart, Best Buy and Sephora, alongside Shop Pay and PayPal for payments. These extend the existing Muse service. Its arrival on glasses remains a plan for the coming months, so the demonstrations should not all be described as universally available features.Connections and timing

Service connections give an agent a way to carry a conversational request into a store and continue the shopping task. For retailers, this could mean a consumer first explains a need to an agent, which then chooses where to look. That gives stores a reason to provide clear product information. The announcement does not supply conversion rates showing how many additional orders those connections generate.

Spotify announced a concrete use case on the same date: after linking an account, Muse can create playlists and schedule playback. Its example uses a road-trip destination to shape a playlist, then starts the music on departure. This is a partner-provided feature example, without typical-user completion rates or measured time savings; it is not an independent field test.Spotify announcement

Glasses reduce description; store coverage shapes the choice

Meta plans for Muse on glasses to receive visual context, including helping someone buy a product they can see. The agent can work in the background and notify them afterward. That could simplify input, but identifying an object and buying the correct variant still involve information that needs checking. A demonstration cannot establish that the entire sequence works reliably every time.Glasses plans

Suppose someone stands in front of a shelf and asks to replenish the same consumable they use at home. Reading the packaging could spare several questions about the product name. Misreading its capacity, specification or pack size, however, could turn faster checkout into more returns. I would have the product confirm differences that change the order, then decide which steps can safely be omitted, rather than asking for confirmation at every stage.

Comparison coverage imposes another limit. Prior context: Axios, September 21 — Amazon blocked Muse shopping.Background Additional stores expand the actionable options without demonstrating that the agent compared every suitable product. That may matter little to someone who has already chosen a brand and model. For someone still deciding what to buy, selection is part of the service itself. I would want recommendations to identify the stores actually checked and any delivery conditions still unverified, so ease of execution and quality of choice can be assessed separately.

Meta initially rolled Muse out to US users; this announcement does not confirm a Taiwan launch schedule.Original availability The available material supplies no total per-transaction costs or incorrect-order rates. A payment provider’s appearance on a partner list does not establish savings for buyers. Users in Taiwan should not treat a global announcement as confirmation that they can already use the service.

I support connecting existing services to agents because it extends stating a need into taking action. For shopping, I would examine how often an order for a specified product needs no subsequent correction, and whether the disclosed comparison scope supports the recommendation. Improvement in the former would support delegating repeat purchases to Muse. If the latter remains incomplete, users should retain room to choose the retailer themselves. The number of connected stores cannot substitute for either result.

The cover reuses this site’s Muse Code brand illustration, not a photograph or product screen from this Connect event. Official and news-image downloads failed because hostnames could not be resolved.

Sources:

Get the latest insights

Join the newsletter to receive my latest articles on GenAI, AI Agents, and architecture.

No spam. Unsubscribe anytime.