Skip to main content

IrisGo

Category: AI Agents

IrisGo is an on-device AI desktop companion that learns user workflows by observation and automates repetitive tasks autonomously, built for AI PCs. IrisGo was founded in 2025. The company is led by Jeffrey Lai. Based in Mountain View, California, United States. Team size: 6-10. Total funding raised: $2.8M. Latest round: Seed. Key investors include AI Fund (Andrew Ng); NVIDIA; Google; Pegatron Ventures.

Founded
2025
Headquarters
Mountain View, California, United States
Team size
6-10
Total funding
$2.8M

Value proposition

IrisGo watches how you work, learns your workflows by observation, and automates them autonomously — no manual configuration or scripting required. It processes data locally on-device for privacy, leverages AI PC hardware (NPU/GPU), and works across all applications on the desktop.

Products and solutions

IrisGo desktop application (Windows & macOS beta); Skills Library (pre-built workflows: email triage, invoice processing, report building, meeting notes, lead scoring, compliance checks, content drafting, onboarding flows, DevOps monitoring, contract summaries, web scraping, etc.); Coding assistant (similar to Codex/Claude Code); Morning Brief Builder; Workflow automation engine with watch-learn-act paradigm

Unique value

The only AI desktop companion that learns workflows by watching (not programming), runs primarily on-device for privacy, and is pre-installed on AI PCs through OEM partnerships — turning the PC into an autonomous workforce.

Target customer

Knowledge workers and white-collar companies; Enterprise teams; Individual professionals; AI PC users

Industries served

Enterprise productivity; Knowledge work; Software development; Sales; Marketing; Finance; HR; DevOps; Legal

Technology advantage

On-device AI processing with local NPU/GPU compute for privacy and speed; Observational learning paradigm (watch-once, automate-forever); Hybrid architecture (local + cloud with end-to-end encryption); Cross-app context awareness; AI PC ecosystem integration (Intel, Acer, Windows); Pre-install distribution moat via OEM partnerships

How they differentiate

Unlike Microsoft Copilot which is tied to Microsoft 365, IrisGo works across all apps on the desktop. Unlike Raycast which requires manual commands, IrisGo learns by watching user behavior. Unlike cloud-dependent agents (Claude, ChatGPT), IrisGo processes data primarily on-device for privacy. Its OEM pre-install strategy with PC manufacturers (Acer, Intel partnership) creates a distribution moat that competitors lack.

Main competitors

Microsoft Copilot (Windows/Office); Raycast; Anthropic Claude (desktop agent capabilities); OpenAI Codex/agents; Apple Siri

Key partnerships

Intel (AI PC ecosystem partner); Acer (OEM pre-install partner); Google for Startups; NVIDIA Inception; Windows (platform); Apple/Mac (coming soon)

Major milestones

2025: Company founded (Irixion Inc.); 2026-01: Raised $2.8M seed round led by AI Fund (Andrew Ng) with NVIDIA and Google participation; 2026-05: Launched beta versions of macOS and Windows apps; 2026-05: Secured OEM pre-install partnership with Acer; 2026-05: Featured in TechCrunch, The Information; Intel partnership for AI PC ecosystem

Growth metrics

Launched beta versions of macOS and Windows apps (2026); Secured OEM pre-install deal with Acer; Featured on TechCrunch, The Information, Tech in Asia

Market positioning

Early-stage startup targeting the emerging AI PC market with a proactive, privacy-first desktop automation agent. Positioned as the "digital butler" for the AI PC era, leveraging OEM distribution partnerships to become the default desktop AI layer.

Geographic focus

United States (headquarters); Global AI PC market via OEM partnerships (Acer, Intel)

About Jeffrey Lai

Ex-Apple Siri Language Engineering Manager (built Chinese language version of Siri); Co-founder & CEO of Tallgeese AI; AI Fund Founder in Residence; B.S. Electrical & Computer Engineering, Carnegie Mellon University

Official website: