Spaces

Product Introduction

1. Definition: Spaces is a desktop-based collaborative AI workspace application (SoftwareApplication/Collaboration) that functions as a unified environment for teams and their dedicated AI agents to manage projects. It operates as a client-side application for macOS and Windows, connecting to external Large Language Model (LLM) APIs like OpenAI's ChatGPT, Anthropic's Claude, and Google's Gemini, or local models.
2. Core Value Proposition: It exists to solve team fragmentation in AI-assisted work by providing a persistent, shared project environment. Its primary value is consolidating team-AI collaboration—including shared chat histories, project files, specialized AI agents, and automated routines—into a single, organized space per project, eliminating the need to context-switch between disparate AI tools and communication platforms.

Main Features

1. Shared Project Spaces: Each project exists as a dedicated space containing all related context. Technically, this space is a synchronized database of conversations, uploaded documents, and agent configurations. For cloud spaces, this data is synced via a secure backend, enabling real-time or near-real-time collaboration. On local-only spaces, data resides solely on the user's machine.
2. Specialized AI Agents with Individual Memory: Users can create multiple AI "specialists" (e.g., a Research Analyst, a Content Editor, a Code Reviewer) within a space. Each agent maintains its own isolated conversation memory and can be configured with custom system prompts, instructions, and connected tools. This allows for role-based, context-aware interactions without cross-contamination of tasks.
3. Scheduled Routines & Playbooks: This feature enables the automation of repetitive workflows. Users can create "routines"—predefined sequences of agent actions—and schedule them to run unattended (e.g., daily, weekly). Playbooks are reusable templates for these routines, allowing teams to standardize processes like generating a morning brief, compiling a weekly sales report, or processing support tickets.
4. Bring-Your-Own-Provider (BYOP) Model: Spaces does not sell AI model access. Instead, it integrates via API with users' existing subscriptions to ChatGPT Plus, Claude Pro, Gemini Advanced, or self-hosted models (like Llama or Mistral via Ollama). This architecture lets users leverage their preferred models and manage costs directly with providers, while Spaces acts as the orchestration and interface layer.
5. Local-First & Hybrid Data Architecture: Sensitive data—including user API keys, connected account credentials, and the core logic of personal agents—is stored locally on each user's desktop. When a cloud space is used, only the necessary project artifacts (chat logs, shared files, routine outputs) are encrypted and synced. This hybrid approach prioritizes data privacy and security.

Problems Solved

1. Pain Point: Disorganized and siloed AI interactions across multiple browser tabs, chat interfaces, and team members, leading to lost context, duplicated effort, and inefficient workflows.
2. Target Audience: Cross-functional product teams (PMs, developers, designers), content agencies, research groups, small business operations teams, and any knowledge-work collective that relies on AI for brainstorming, drafting, analysis, and automation.
3. Use Cases:
   • Product Development: A shared space where PMs, developers, and QA can use AI to brainstorm features, review code snippets, and document decisions in a continuous thread.
   • Content Production: An editorial team collaborating with AI writers, editors, and SEO specialists on a content calendar, with all drafts, feedback, and revisions in one place.
   • Operational Automation: A small business owner setting up scheduled routines for an AI agent to analyze daily sales data, summarize customer support emails, and post a digest to a team channel.

Unique Advantages

1. Differentiation: Unlike individual AI chat apps (ChatGPT web interface) or single-user desktop assistants, Spaces is built for persistent, multi-user, multi-agent collaboration. Compared to project management tools, it deeply integrates AI as a core, actionable team member rather than just a plugin.
2. Key Innovation: The "space-per-project" model that unifies human team chat, AI agent interactions, project files, and automation into a single, persistent context. The hybrid local/cloud architecture that keeps sensitive credentials local while enabling team sync is a significant technical and privacy-focused design choice.

Frequently Asked Questions (FAQ)

1. Is the Spaces desktop app really free? Yes, the core desktop application for individual use on your local machine is completely free and does not require creating an account. You only incur costs if you need to create a cloud-based space to collaborate with other team members in real-time.
2. How does billing work for team collaboration in Spaces? Collaboration requires a paid "Spaces Cloud" subscription, billed per seat. Each member who needs to join and actively work within a shared cloud space requires their own license ($1.49/month or $10.99/year). This model is based on the principle that each user brings their own configured AI agents to the shared environment.
3. Where is my API key and chat data stored for security? Your LLM provider API keys and the core configuration of your personal AI agents are stored locally on your computer and are never transmitted to Spaces' servers. In a cloud space, only the project content you choose to share (conversation history, uploaded files) is encrypted and synced.
4. Can I use multiple AI models like ChatGPT and Claude simultaneously in Spaces? Absolutely. You can connect multiple LLM provider accounts (e.g., both OpenAI and Anthropic) to Spaces. You can then assign different models to different AI specialists within the same project, allowing you to leverage the specific strengths of each model for different tasks.
5. What are the system requirements for running the Spaces desktop app? Spaces is a native desktop application available for computers running macOS (with Apple Silicon or Intel processors) and Windows (10 or 11). It requires a standard internet connection for initial setup, model API calls, and cloud space syncing.