One More Note on Agents, Meta Connect, Meta Enterprise Platform (Stratechery Update, 2026-09-29)

Why this is in the vault

Direct sequel to yesterday's "Apps, Agents, and Aggregation" — Thompson adds a consumer-adoption coda, recaps Meta Connect hardware, then argues Meta's brand-new Meta Enterprise Platform (announced the same morning) is a strategic mistake that risks the consumer agent lead Muse just built.

Mapping against Ray Data Co

Mapping strength: medium. The concrete anchor is the inverse of yesterday's note: Thompson's warning that "serving enterprise customers is completely different than serving consumers... there are very few companies who are good at both" is the same structural claim that underwrites RDCO's own posture of staying inside an incumbent agent's enterprise distribution (Claude via phData) rather than trying to be a consumer-facing agent product, per [[project_credibility_for_phdata_sales]]. Read as a caution in the other direction: if a consumer-native winner like Meta struggles to bolt on enterprise go-to-market, the enterprise-native lane (where phData sits) is the harder-to-contest side of the split, not a lesser one.

Thompson's identity argument — one Facebook profile forcing "all is revealed" was a mistake, and Copilot's personal/professional split confirms it — is a sharper version of the same point implicit in Ray's own Mac Mini setup: work (phData, RDCO) and personal (family, health, finances) run through the same always-on agent but stay in separate channels and vault sections, not a single undifferentiated identity. Worth a gut-check next time an agent-identity or persona-merging decision comes up.

Contrarian check: Thompson's Muse-for-Small-Business allowance ("start there... going for enterprise now is a mistake") implicitly validates a small-business/solo-operator wedge as the sane middle step between consumer and enterprise — closer to RDCO's own actual scale than either extreme. Don't over-read this as blanket enterprise-vs-consumer dogma; it's a sequencing argument, not a permanent one.

The core argument

• Consumer-agent adoption coda: normal people don't care about "productivity," but current app/web UX is still hard; once ordinary users figure out agents (not yet), they may love them more than technical users do, and only then do the strategic Aggregation stakes from yesterday's piece actually bite.
• Meta Connect recap (via Bloomberg/Gurman): $1,299 Meta VR Glasses (externalized compute pack, ~100g, ~1/6 the weight of Vision Pro) undercut Apple on price and comfort; camera-free Ray-Ban Meta Audio glasses address the privacy backlash while adding battery life; the palm-sized "Muse Charm" gadget (Tamagotchi-like, built by ex-Apple design chief Alan Dye's lab) signals Meta's devices org now has a single north star — all hardware is a Muse delivery vehicle, justifying over a decade of twelve-figure Reality Labs losses if Muse is a hit.
• Meta Enterprise Platform reaction: Meta is bundling Muse, a coding tool, model access, and business agents into a new enterprise sales motion under ex-MongoDB CEO CJ Desai, reporting to Zuckerberg. Thompson calls this a mistake — enterprise and consumer require opposed go-to-market and product disciplines (Google Cloud under Kurian took over a decade to reach a weak third place, arguably at the cost of Google's consumer AI focus).
• Identity argument: revisits his own "Social Networking 2.0" thesis against Zuckerberg's "one identity" doctrine (from The Facebook Effect) — separate private/professional identities are a superior product experience, not a lack of integrity, and Microsoft already learned this by splitting personal/professional Copilot. Merging work into Muse's cute, fun positioning cuts against that.
• Verdict: Meta should stay focused on owning the consumer agent space (Muse for Small Business is an acceptable adjacent step); chasing enterprise now, likely as a rationalization for its >$100B AI infrastructure spend, is premature and risks the larger prize.

Sponsorship and bias

No sponsor block — Stratechery is Thompson's own subscription-funded product with no third-party advertising; the footer here is standard renewal/subscription-management boilerplate only. Bias to flag: Thompson is an open Muse enthusiast mid-cycle on a three-week "Meta momentum" thesis (explicitly referenced from his own Sharp Tech and Dithering appearances this same week), so the framing of Meta's enterprise move as a "mistake" is argued from a position already invested in the consumer-Muse-wins narrative; the Google Cloud comparison is also somewhat selective (ignores that Google Cloud is profitable and strategically valuable despite third place).

Related

• [[2026-09-28-stratechery-apps-agents-aggregation-inspiration]] — the article this Update directly follows up on and extends
• [[2026-09-22-stratechery-amazon-blocks-muse-aggregator-moat]] — the Meta-Amazon Muse dispute referenced via this week's Sharp Tech episode
• [[2026-09-28-ark-invest-amazon-blocks-muse-agentic-commerce]] — independent same-week coverage of the Meta-Amazon Muse friction
• [[2026-08-06-alphasignal-prime-intellect-meta-muse-skills-v1.2]] — Muse platform build-out, cross-source Muse tracking
• [[project_credibility_for_phdata_sales]]
• [[project_l5_north_star_strategic_direction]]