CHAPTERS
- 0:00 – 0:31
Personal agent fatigue sets the bar: “exceptional or I’m out”
Claire opens with mixed feelings: Claude Duet looks great, but she’s overwhelmed by the flood of personal agents. She frames the review around one question: can Duet be exceptional enough to earn a top spot in her stack?
- •Anthropic releases Claude Duet, a new personal agent
- •She’s saturated with agents and skeptical of yet another option
- •To win attention, the product must be “exceptional,” not merely good
- •The episode will test whether Duet makes the cut
- 0:31 – 1:32
What Claude Duet is: a mobile-first Claude-powered agent for SMBs
She quickly explains what Duet is and who it targets. It’s an iOS/Android personal agent synced with Claude, designed for busy professionals and small business owners rather than power users running large agent fleets.
- •Mobile app (iPhone/Android) connected to Claude
- •Uses Claude tools/plugins/connectors
- •Positioned for SMB owners and busy professionals (PM/CEO)
- •Meant to handle everyday “coworker” tasks from your phone
- 1:32 – 2:32
Design polish and onboarding: where Anthropic beats competitors
Claire praises Duet’s UI/UX: clean, thoughtful, and trustworthy from the first run. She contrasts this favorably with OpenAI’s Dots, emphasizing how much smooth onboarding matters for adoption.
- •Exceptional design polish; “no sharp edges”
- •Clear onboarding: what it does, what it can access, how to start
- •Feels trustworthy and easy to get running immediately
- •Strong contrast: Duet polished vs. Dots’ rougher launch
- 2:32 – 4:33
Core workflow primitives: connectors, task lists, artifacts, and channels
Duet’s core building blocks are designed to make agent work trackable and organized. Claire highlights inherited connectors, task lists for goals, first-class artifacts, and Slack-like channels to segment work streams.
- •Inherits Claude connectors (Gmail, Google, Figma, etc.)
- •Task lists help track what you vs. the agent are doing
- •Artifacts are first-class; it can read existing Claude artifacts
- •Channels model sub-chats like Slack workspaces (e.g., a sales channel)
- 4:33 – 5:03
Even the icon matters: brand-consistent and unobtrusive
She calls out a small but telling detail: the icon design. It’s cute, abstract, and “Claude-y” without being distracting—another sign of product maturity.
- •Icon feels on-brand and thoughtfully designed
- •Non-offensive and unobtrusive compared with more character-driven agents
- •Signals care in product aesthetics
- •Design details reinforce perceived quality
- 5:03 – 7:05
Big miss #1: not proactive enough to pull her back in
Despite being positioned as proactive, Duet didn’t re-engage her after initial setup. Claire argues that calmness and subtlety can hurt retention when users juggle many agents.
- •After initial tasks, Duet didn’t nudge her to return
- •Notifications weren’t “noisy” enough to stay top-of-mind
- •Generic morning briefings aren’t compelling anymore
- •She wants stronger re-engagement loops to compete with Grok Bot and Muse
- 7:05 – 7:35
Testing real work: email/Slack triage and shopping Warriors tickets
Claire puts Duet to practical tests—triaging messages and shopping for event tickets. It performs “okay” but not exceptionally, with issues around prioritization and follow-through.
- •Email/Slack scan delivered mediocre prioritization/context
- •Ticket-shopping used as a practical agent evaluation
- •Quality felt acceptable but not standout
- •Lack of strong closure on tasks reduces trust and usefulness
- 7:35 – 10:07
Why latency is a deal-breaker for personal agents
She argues speed is a core feature for agents meant to feel like responsive coworkers. Even with strong models (e.g., Opus), long waits and slow threads cause her to disengage and forget tasks in progress.
- •Claude’s slowness makes synchronous collaboration painful
- •Fast agents (Muse, ultrafast models) keep her engaged
- •Long task runtimes (10+ minutes) break momentum
- •Slow loops + quiet notifications = forgotten threads and dropped workflows
- 10:07 – 12:07
Missing desktop app: mobile-only breaks daily work habits
Claire stresses that serious work happens at a computer, not on a phone. Without a desktop presence, Duet can’t sit in her workflow the way Grok Bot, Muse, and even Dots can.
- •She doesn’t want to work from phone during the day
- •Desktop app provides persistent presence and reminders
- •Grok Bot/Muse succeed partly due to desktop agents
- •Duet needs cross-device continuity to become a daily driver
- 12:07 – 13:08
What it does well: sales pipeline review and on-the-go deck creation
Duet shines when connected to business systems: it can pull pipeline context from tools and generate a usable Google Docs deck from a phone. However, it doesn’t meet Claire’s higher design bar (e.g., Figma-level fidelity).
- •Strong value when already connected to Slack/email/CRM
- •Can summarize pipeline context quickly once connected
- •Creates decks in Google Docs from mobile—useful for many SMB users
- •Design output doesn’t match her preferred brand/design rigor
- 13:08 – 13:38
Launch videos and the “local skills” portability gap
Claire tries to replicate her Claude-based launch video workflow on Duet. She finds Duet can’t automatically access her local skills, highlighting a broader ecosystem problem: inconsistent context and capability portability across surfaces.
- •Attempted to generate launch videos from Duet
- •Duet doesn’t ingest local skills by default
- •May require turning local skills into Claude plugins
- •Broader challenge: what context/tools port cleanly across products
- 13:38 – 15:09
Single-player vs multiplayer agents: collaboration still lacking
She notes Duet is built for solo use, while her work increasingly depends on team workflows. Without easy sharing and “multiplayer” agent experiences (like Slack-based team agents), Duet may not embed into her company’s stack.
- •Duet feels optimized for individual workflows
- •Claude ecosystem hasn’t fully “cracked” team collaboration
- •She’s shifting Grok Bot usage toward shared Slack/team agents
- •Even small teams need sharing to justify adopting a new agent tool
- 15:09 – 16:21
Final verdict: beautifully built, but it doesn’t earn a switch
Claire recommends trying Duet, praising its polish and elegance. But for her, the combination of low proactivity, high latency, lack of desktop, and limited collaboration keeps it from displacing her current top agents.
- •~100x more polished than Dots at launch (in her view)
- •Good models and strong foundations (artifacts/channels)
- •Needs improvements: urgency/proactivity, speed, desktop presence, collaboration
- •She’s unsure it will become part of her personal stack
