Vision
Stardust aims to be the operating system for getting work done—a platform where humans and AI agents collaborate seamlessly on complex projects.This roadmap represents our current thinking and may evolve based on user feedback and technological advances.
Current Focus
What We’re Building Now
Multi-Agent Coordination
Advanced swarm capabilities for parallel development work
Real-Time Multiplayer
Google Docs-style collaboration with presence and cursors
Issue Tracker Integration
Deep integration between task management and AI workflows
Desktop Apps
Native applications for macOS, Windows, and Linux
Horizon 1: Foundation (Next 6-12 Months)
Multiplayer Excellence
- Live collaboration: Real-time cursors, presence, and editing
- AI in channels: Agents as first-class team participants
- Resumable streaming: LLM responses that survive disconnections
- Conflict resolution: CRDT-based merging for concurrent edits
Mobile Experience
- Native iOS app: Full-featured mobile client
- Native Android app: Parity with iOS
- Offline support: Work without connectivity, sync when online
- Voice capture: Quick task entry via voice
Agent Capabilities
- Swarm orchestration: Coordinate multiple agents efficiently
- VCS-native operations: Every change tracked in Git
- Long-range memory: Sessions that span days without context loss
- Error recovery: Bisection and rollback for agent mistakes
Horizon 2: Intelligence (12-24 Months)
Contextual Intelligence
- Predictive prioritization: AI learns your patterns over time
- Smart scheduling: Optimal task ordering based on energy and context
- Proactive alerts: Surface issues before they become problems
- Cross-project insights: Patterns across all your work
Calendar Integration
- Native calendar sync: Two-way sync with Google, Outlook, Apple
- Time blocking: Automatically schedule focused work time
- Meeting intelligence: Prep materials and follow-up tasks from meetings
- Deadline awareness: Priority adjustments based on calendar
Advanced AI Features
- Multi-model routing: Right model for each task type
- Custom fine-tuning: Agents that know your codebase deeply
- Evaluation framework: Measure and improve AI quality
- Prompt library: Reusable, tested prompt templates
Horizon 3: Platform (24-36 Months)
Agent Marketplace
- Custom agents: Build and share specialized agents
- Agent templates: Quick-start agents for common tasks
- Third-party integrations: Agents that connect to external services
- Revenue sharing: Monetize your agent creations
Multi-Agent Orchestration
- Complex workflows: Multi-step agent pipelines
- Human-in-the-loop: Approval gates and review points
- Resource management: Optimal allocation of compute
- Quality gates: Automated quality checks between steps
Enterprise Features
- Team analytics: Insights into team productivity
- Custom workflows: Configurable task lifecycles
- Audit logging: Compliance-ready activity tracking
- SSO/SAML: Enterprise authentication
Horizon 4: Scale (36-48 Months)
Self-Hosted Deployment
- On-premise: Run Stardust in your own infrastructure
- Air-gapped: Fully offline deployments
- Custom models: Bring your own AI models
- Data sovereignty: Complete control over your data
Global Scale
- Multi-region: Data residency options
- High availability: 99.99% uptime SLA
- Enterprise SLA: Dedicated support and guarantees
What We’re Not Building
Staying focused means saying no. These are explicit non-goals:- Project management theater: No Gantt charts, burndown charts, or velocity tracking
- Engagement optimization: We want you to use Stardust less, not more
- Social features: No likes, followers, or activity feeds
- Complexity for complexity’s sake: If it doesn’t reduce cognitive load, it doesn’t ship
How We Prioritize
Our prioritization framework:- Does it reduce cognitive load? If not, skip it.
- Does it help users ship faster? Bias toward action.
- Does it compound over time? Prefer investments that get better.
- Can we ship it quickly? Small bets, fast iteration.
Get Involved
Share Feedback
- GitHub Issues: Report bugs and request features
- Community Discord: Join discussions with other users
- Email: feedback@stardust.computer
Contribute
- Open Source: Stardust is open source—contributions welcome
- Documentation: Help improve these docs
- Integrations: Build integrations with other tools