> ## Content Index > Fetch the complete content index at: https://spyglass.org/llms.txt > Use this file to discover other available public pages before exploring further. # The AI Assistant Wars - URL: https://spyglass.org/the-ai-assistant-wars/ - Published: 2026-10-15T22:37:00.000Z - Updated: 2026-10-16T11:38:46.000Z - Description: A chat about the state of AI personal assistants and if this is really the time for them to shine... - Author: M.G. Siegler - Tags: big technology, ai, muse, meta, apple, siri ai, openai, chatgpt, instinct, microsoft, copilot, anthropic, claude, google, gemini, tech On September 1, [I wrote a column](https://spyglass.org/the-swiss-assistant/) looking back at the history of AI "personal assistants" and wondering if now might be the time, thanks largely to the rise of LLMs, for them to actually take over. While [Instinct](https://instinct.com/?ref=spyglass.org) and [Town](https://www.town.com/?ref=spyglass.org) were getting some early love (months after the initial OpenClaw love [predictably died down](https://spyglass.org/openclaw-versus-closed-claude/)), certainly on VC Twitter, the big players were obviously looming... Fast forward six weeks and now it's a full-on war in the space. To the point where my computer while recording [this *Big Technology Podcast*](https://www.youtube.com/watch?v=fq60laurHBw&ref=spyglass.org) with Alex Kantrowitz was on the verge of melting down from running various agents in the background. During the chat, we dive into why perhaps these services haven't taken off in the past (we're approaching the 40th anniversary of [the old Apple "Knowledge Navigator" concept video](https://www.youtube.com/watch?v=umJsITGzXd0&xstg=CAMSEBUJ%5Fb-oH-PhF0yjBgb%5FzzE%3D&ref=spyglass.org)) and what's different now. Namely, I think the proactive nature of these services helps to overcome a real "blank page meets boil the ocean" issue. That plus leveraging the tried-and-true messaging paradigms (bolstered by ChatGPT itself), should get people to at least *try* these products. But to keep people sticking around, there will need to be actual ongoing utility – and saving money is a good one! Granted, it may destroy the banking industry and/or anything that relies on ongoing subscriptions, but still... So *if* these services take off this time, it may be more a question of timing – how long will all of this take to go truly mainstream? This is obviously where Muse has a huge advantage given the ability to leverage Meta's other multi-billion-user products. And yeah, it remains the number one overall app in the App Store [a month after launch](https://spyglass.org/metas-amusing-muse/). Sora, by comparison, could only hold the top of the store for a few days, [even with](https://spyglass.org/openai-sora-2-app/) OpenAI pushing it via ChatGPT. This is a big part of why I think Muse [has a shot here](https://spyglass.org/meta-muse-connect-keynote/) – well that and the fact that it's good. If Meta can convince users to overcome their privacy fears – something which the product is rightly focused on from the get-go. Another interesting timing wrinkle: if the current crop of movies where these [major tech players are villains](https://spyglass.org/tech-hollywood-villain/) doesn't poison this well, as it were... I almost can't believe I'm the one defending [the blood-sucking social network](https://spyglass.org/metas-amusing-muse/) here, but the fact that Muse is this good without a true frontier model powering it is impressive – and points to a world where it gets even better, fast with "Watermelon" (which, of course, is now live, just after our conversation, [in the form of "Wonder](https://spyglass.org/thoughts-on-meta-back-at-the-ai-frontier-sora-as-a-startup/)). Anyway, Alex is more Team Dots. That product is also well done, but there I worry about OpenAI's [focus](https://spyglass.org/theres-always-a-siren/) and [muddled messaging](https://spyglass.org/openai-dots-chatgpt-ai-assistant/). But it certainly helps, distribution-wise, that it's baked right into ChatGPT itself. Though not nearly as much as the boost Siri AI naturally gets from being pre/auto-installed on basically *all* Apple devices in the wild. That's why Alex has Apple as the dark horse here, and it's hard to disagree with that, [as I've written about in the past](https://spyglass.org/siri-ai/). Still, Siri AI, which, while better than Siri, is no Muse or Dots or the like. Perhaps that points to Apple buying Instinct... What about Microsoft? [Their](https://spyglass.org/copilot-ai-super-app/) "super app" is [here](https://spyglass.org/inklings-metas-vision-pro-copilots-super-app-streaming-ads-doj-helps-apple/#:~:text=%F0%9F%9B%A9%EF%B8%8F%20The%20Copilot%20%22Super%20App%22%20Lands) with (the actually smartly named) "Autopilot" in the passenger seat. Can their hold in enterprise translate to consumer success here too? Or will [Bring Your Own AI](https://spyglass.org/bring-your-own-ai/) become the norm at work? Um, where's Anthropic in this personal AI assistant race? Are they thinking that Claude Code/Cowork will morph into their play? Or do they still simply not care about the consumer market? What about Amazon? What are they up to here beyond [blocking](https://spyglass.org/thoughts-on-amazon-blocks-metas-muse-apples-iphone-duo-pencil-pivot/) other company's agents? Is [this whole thing](https://spyglass.org/thoughts-on-the-great-ai-agent-race-a-super-ai-rebrand-attempt/) just going to quickly devolve into another turf war? [But really, where is Google?!](https://spyglass.org/why-didnt-google-build-muse/) They *clearly* should have been the one to build Muse, as the comments around my original post can attest. And yeah, they've been seemingly working on it with 'Gemini Spark', but even in beta, it's just not breaking through in the same way. Part of the problem is clearly some [internal](https://spyglass.org/google-deepmind-hassabis-dean-gemini/) [chaos](https://spyglass.org/google-gemini-fiasco/) and [confusion](https://spyglass.org/inklings-spring-loaded-ternus-endless-geminis-mathpocalypse-openais-missing-arr-ok-googlebooks/#:~:text=%F0%9F%92%AB%20Google%20Announces%20Their%2016th%20Agent%20and%2012th%20Gemini) which led to a [Google I/O conference](https://spyglass.org/google-io-2026/) with some vaporware announcements – notably, Gemini 3.5 Pro. But now that Gemini 4 is [here](https://spyglass.org/inklings-apples-homehub-a-nimble-apple-spacex-phone-mountdancewarbrodisco-argons-conquest-flight-club/#:~:text=%F0%9F%90%89%20Gemini%204%20Argon%27s%20Conquest%3F), maybe Google is [finally](https://spyglass.org/predicting-googles-ai-earthquake/) ready to reorient themselves again and come storming back in AI – *again*?