jo

What do they actually do

jo is a macOS assistant (Apple Silicon only) in gated preview that you invoke by voice (“hey jo”) or keyboard (⌘-J). It accepts speech or typed input, speaks and types responses, and provides a live, three‑point summary of the frontmost window that updates as you switch pages or documents. Access is via waitlist invites the team is sending out gradually, not an open launch yet (launch blog, YC profile).

Beyond the Live Summary, jo can chain practical tasks: it can run and pre‑fill searches (e.g., flights, listings, videos), help locate a past email, draft an email, or open apps/pages to get you most of the way to an outcome. It also syncs with Google Calendar to find and group open slots to protect focus time. The product emphasizes incremental trust controls (opt‑in mic, explicit Live Summary sharing, short remote caches) and keeps some data local while using cloud models when needed (launch blog).

Under the hood, jo uses local small models for low‑latency Live Summary and remote models hosted on Azure for heavier tasks; it currently relies on third‑party audio transcription/synthesis. It works on M‑series Macs (M2+ recommended) and is free during the preview, with pricing to be announced in early 2025. The team signals upcoming mobile apps and broader integrations, with a longer‑term goal of moving from a helpful assistant toward a trusted agent as safety patterns are proven (launch blog, roadmap).

Who are their target customer(s)

• Knowledge workers who read long articles/reports/docs: They spend too much time skimming and lose context when switching windows; jo’s three‑point Live Summary helps them get the gist quickly (launch blog).
• Meeting‑heavy professionals (PMs, execs, recruiters): They waste time checking calendars and proposing times; jo syncs with Google Calendar and suggests grouped slots to preserve focus blocks (launch blog).
• People who assemble scattered info (researchers, lawyers, customer‑facing reps): They struggle to locate past emails or assemble relevant results; jo chains searches and pre‑fills results to get them most of the way faster (launch blog).
• Mac power users who prefer hands‑free, in‑flow control: Context switching between mouse/keyboard breaks flow; jo is voice‑first (or keyboard‑invoked) and responds in voice/text while following the frontmost window (launch blog, YC profile).
• Privacy‑sensitive professionals or IT evaluators: Uncertainty about what data leaves the device; jo emphasizes local summaries, a hybrid local/remote approach, and explicit data controls (launch blog).

How would they acquire their first 10, 50, and 100 customers

• First 10: Pull from the waitlist and founder/YC network with prioritized invites and brief 1:1 onboarding to watch real workflows, gather quotes, and offer a direct support channel. Allow each early user to invite one teammate for faster feedback loops.
• First 50: Expand invites to Mac power‑user and meeting‑heavy communities via live demos and offer a free month of team seats in exchange for structured feedback and example workflows that become onboarding playbooks.
• First 100: Add a simple referral to skip the waitlist, publish short how‑to videos/templates, and run 2–3 small‑team pilots with clear privacy/security docs and a named contact. Use pilot metrics and quotes to drive broader adoption.

What is the rough total addressable market

Top-down context:

Initial TAM is the set of Apple Silicon Mac knowledge workers willing to pay for a desktop AI assistant that summarizes on‑screen content and helps with scheduling and search—an early wedge within the broader AI productivity/assistant market.

Bottom-up calculation:

If 250k–500k Mac professionals adopt at $20/user/month, that implies ~$60M–$120M in annualized revenue potential for the early market wedge. Larger TAM expands with Windows/mobile and enterprise rollouts.

Assumptions:

• Apple Silicon Mac knowledge workers in core markets number in the low millions; 5–10% are near‑term adopters (≈250k–500k).
• Willingness to pay averages ~$20/user/month for a real‑time assistant delivering daily value.
• Early sales are mostly individual/small‑team seats; enterprise ramps later.

Who are some of their notable competitors

• Rewind: Privacy‑first “record everything” memory app for macOS/iOS that stores searchable screen/audio locally. Overlaps on context and summaries but focuses on continuous, passively captured memory vs. jo’s live front‑window summary and action chaining (Rewind).
• Raycast: Keyboard‑first Mac launcher with AI chat/summarization and an extension ecosystem. Similar in staying in‑flow with quick actions; different in being a command/launcher platform, not a voice‑first assistant that follows the active window or syncs calendars by default (Raycast AI).
• Otter.ai: AI meeting notetaker for transcription, summaries, and action items. Overlaps on meeting capture/summaries; Otter is meeting‑centric and collaboration‑oriented vs. jo’s per‑window live summaries and desktop task chaining (Otter features).
• Microsoft 365 Copilot (Outlook/Teams/Calendar): AI features inside Microsoft 365 that summarize threads, draft emails, and suggest meeting times. Strong in Microsoft‑native workflows; jo is Mac‑native, voice‑first, and follows any frontmost window across apps (Copilot docs).
• Superhuman: Email client with fast scheduling and AI drafting. Competes for meeting‑heavy pros on scheduling; Superhuman is inbox‑centric, while jo is a system‑level assistant with live window summaries and cross‑app task chains (Superhuman).