Who Wins The AI Assistant Wars? + Why Didn’t Google Build…

PodcastBig Technology PodcastM.G. SieglerOct 5

The discussion examines how AI assistants could gain traction through familiar interfaces, proactive help, and services such as bill review, while raising questions about trust, privacy, and subscription churn. It compares Meta, OpenAI, Microsoft, Google, and Apple on product execution and distribution, with cloud-versus-desktop capabilities emerging as another competitive factor. Separate discussion considers banks’ exposure if agents move customer cash, along with Instinct’s rapid fundraising and possible acquisition interest.

Artificial Intelligence Technology Stocks Banking

Preview · 23 more points in the app

Read the Full Summary

Key Takeaways

The guest argues that LLMs have renewed the push for consumer AI assistants, while familiar messaging services such as WhatsApp could lower adoption friction. Proactive prompts—like handling an inbox message or setting up recurring bill payments—could also help users get past the blank-page problem.

The guest sees AI agents as potential tools for finding household savings by reviewing bills and canceling unused subscriptions, including HBO Max or Netflix. The host also raises a related risk for banks: if agents move cash out of non-interest-bearing accounts, the industry could take a substantial but unquantified hit.

The guest sees Instinct’s independence from major tech companies as a possible advantage amid information silos, but says the startup still depends on outside models and cloud providers and lacks large competitors’ distribution. Assistants may also need access to Gmail and Google Calendar, while users’ reluctance to share those services with Meta illustrates the trust barrier to connected assistants.

The guest describes Meta’s Muse as consumer-focused, Microsoft’s Copilot as more enterprise-oriented, and OpenAI as pursuing a hybrid strategy, while noting uncertainty about whether people will use separate work and home assistants. The speakers cite existing reach—including Meta’s multibillion-user base and OpenAI’s stated 1.2 billion weekly active users—as a distribution advantage; the host predicts OpenAI will win and says Apple could be a dark horse because its AI could reach billions of devices.

The guest says Muse’s cloud computer lets users observe browsing, take control, and access the virtual machine, though security concerns remain. He contrasts that cloud model with desktop workflows involving local files, such as an Excel sheet outside the cloud, and says the balance between cloud and desktop capabilities could help determine which services win.

TECHNOLOGY

LLM Assistants — The guest argues that widely available LLMs have made consumer AI assistants newly feasible, renewing the push to build a personal assistant.

“And, you know, in some ways it is, you know, analogous to what we have right now, which is interesting. But obviously, was way too early…”—

Familiar Interfaces — Instinct aims to reduce adoption friction by working through messaging tools people already use, rather than requiring a separate app. Meta’s Muse also launched with WhatsApp integration.

“Instinct, I think, wisely recognized that they could sort of get off the ground running fast by simply leveraging already existing messaging paradigms…”—

Proactive Prompts — Proactive prompts could address the blank-page problem by surfacing pending tasks rather than waiting for users to formulate requests. Examples include replying to an inbox message and setting up recurring bill payments.

“Now I feel like you get around that problem again with some level of prompting…”—

ECONOMY

Deposit Vulnerability — The host relays bankers’ concern that the banking industry depends on customers keeping money in non-interest-bearing accounts. He raises the possibility that agents moving cash into other accounts could cause a substantial financial hit to the industry, though the impact is not quantified.

“By the way, there's been some…”—

MARKETS

VC Deal Access — MG Siegler suggests some VCs may promote Instinct on social media partly to improve their chances of accessing a future deal. Public enthusiasm may therefore reflect investment motives as well as interest in the product.

“Like, VCs feel like they gotta give some love on social media so that they have any shot of, of potentially getting their foot in the door if there happens…”—

Personal Assistant Surge

Siegler revisits his view that no AI assistant had nailed the experience, arguing that new models and familiar messaging interfaces have made the category newly viable.

Instinct’s Proactive Interface

Instinct lowers the barrier to trying an assistant by working through existing chat services and proactively suggesting useful tasks rather than leaving users with a blank prompt.

Bills, Savings, And Churn

The speakers point to bill review and savings as compelling assistant use cases, while warning that agents could prompt customers to cancel subscriptions and move cash.

Banking And Subscription Risk

Agents that optimize balances, rewards, and recurring subscriptions could disrupt banks and subscription businesses that depend on customer inertia.

Assistant Market Contenders

The discussion compares Instinct, Meta, OpenAI, Microsoft, and other entrants, emphasizing the startups’ neutrality but also their limits in funding, infrastructure, and distribution.

Meta Muse’s Early Execution

Amazon’s restrictions illustrate how platform conflicts could constrain assistants, while Siegler says Meta’s standalone Muse product is a surprisingly polished early release.

Consumer Versus Enterprise

Meta appears focused on consumers, Microsoft on enterprise, and OpenAI between the two; the eventual winner may need to reconcile work and personal assistance.

Cloud And Desktop Control

Muse’s dedicated cloud computer can ease some security concerns, but desktop access remains important for tasks involving local files and established workflows.

Distribution And User Trust

Siegler sees first-mover reach and integrations as major advantages, while the speakers discuss how privacy concerns and public perceptions could affect adoption.

Integrations Shape Winners

A large installed base may favor big technology companies, but assistants also need durable access to email, calendars, messaging, and other services.

Anthropic’s Assistant Strategy

Siegler suggests Anthropic may extend Claude Code and Cowork toward assistance, but says the company has historically leaned toward professional productivity over consumer products.

Meta, OpenAI, And Apple

The host favors OpenAI’s product experience and scale, while Siegler highlights Meta’s models and Apple’s device integration as potential competitive advantages.

Apple’s Siri Trust Deficit

Apple has broad access to device context and services, but users may forget or distrust Siri because of its long history of underwhelming performance.

Google’s Assistant Gap

Siegler argues Google has the services and talent to compete but has struggled to ship a compelling, coherent assistant product on time.

Google’s Organizational Friction

Google’s many teams and overlapping assistant efforts can confuse users and hinder execution, despite leadership attempts to unify its AI organization.

Acquisition And Rivalry Outlook

Siegler expects acquisition interest in assistant startups and notes that Amazon, xAI, Microsoft, and other rivals remain part of an unsettled competitive landscape.