Back to Docs

Fluid Software Philosophy

Democratizing software development through dynamic, intent-driven platforms that evolve with their users

Our Vision

A self-evolving AI platform where the conversational AI orchestrates all services, users can directly interact with the same services, the platform uses itself to build itself, and everything is transparent, customizable, and self-improving.

What Makes Software "Fluid"?

Fluid software provides composable building blocks that users and AI can orchestrate together—files, databases, workflows, automations—creating emergent capabilities through intelligent coordination.

Rather than generating code from scratch, Awareness provides powerful managers for files, data stores, workflows, and UI components. The AI orchestrates these real systems through natural language, enabling collaboration where users and AI share the same tools and capabilities.

📁

Real File Management

Upload, version, search, and organize files in MinIO. AI can read, analyze, and extract insights from your actual documents—not simulated data.

🗄️

Database Orchestration

Connect PostgreSQL, Neo4j, or document stores. AI translates natural language to SQL, queries your data, and helps you explore connections.

⚙️

Workflow Automation

Create multi-step workflows with conditions, loops, and human-in-the-loop approvals. AI can trigger, monitor, and adapt workflows based on context.

🤝

Collaboration & Sharing

Spaces provide multi-tenant isolation. Share files, data, and workflows with team members. AI agents respect the same permissions users have.

🔌

OAuth Connections

Connect external services (Google, Slack, GitHub) via OAuth. AI can use these connections to fetch data, send messages, or trigger actions.

Emergence: When Simple Systems Create Complex Behavior

Emergence is when simple components, following simple rules, produce surprisingly complex and intelligent behavior through their interactions. In Awareness, emergence happens when AI agents coordinate across files, databases, workflows, and memory—creating solutions neither the user nor system designer explicitly programmed.

🧩Simple Components

  • • File manager: read, write, search
  • • Database adapter: query, insert, update
  • • Workflow engine: sequence, condition, loop
  • • Memory system: store, retrieve, consolidate
  • • Tool registry: define, execute, approve

Emergent Capabilities

  • • AI discovers patterns across files and databases
  • • Workflows adapt based on learned context
  • • Memory informs file organization strategies
  • • Cross-domain insights from tool coordination
  • • Self-improving automation through reflection

🔬Example: Emergent Data Pipeline

User asks: "Analyze customer feedback and update our roadmap"

AI coordinates:

  1. 1. Searches files for feedback documents (File Manager)
  2. 2. Extracts sentiment and themes (LLM + Memory)
  3. 3. Queries database for existing feature requests (Data Store)
  4. 4. Matches feedback to requests (Vector Search + Neo4j)
  5. 5. Triggers workflow to notify product team (Process Engine)
  6. 6. Updates roadmap document (File Manager + Approval)

Emergent result: An end-to-end analytics pipeline the system wasn't explicitly programmed to perform, created by intelligently coordinating simple components.

Why Emergence Matters

  • Unpredictable Innovation: Solutions emerge that designers didn't anticipate
  • Scalable Complexity: Simple rules scale to handle complex scenarios
  • Adaptive Behavior: System responds to novel situations without explicit programming
  • User Empowerment: Non-technical users get expert-level system orchestration

Five Foundational Principles

1
🧱

Composable Building Blocks

Powerful, reusable managers for files, databases, workflows, and UI components. Both users and AI orchestrate these real systems through shared interfaces.

Examples:
  • File Manager handles local, S3, and Git storage
  • Data Store Manager supports PostgreSQL, Neo4j, document stores
  • Workflow Engine executes multi-step automations with human approvals
  • UI components render dashboards from structured definitions
2
🤝

AI + Human Parity

Anything the user can do, the LLM can do (and vice versa). Both have equal capability to create and modify systems through the same interfaces.

Examples:
  • Same APIs for user actions and AI actions
  • Tool system accessible to both
  • Shared context and permissions
3
🏠

Self-Hosted First

Full control, no vendor lock-in, privacy by design. Empowering users with complete ownership of their platform and data.

Examples:
  • Deploy on your own infrastructure
  • No dependency on cloud vendors
  • Complete data sovereignty
4
🌍

Cooperative Ownership

Built for the community, owned by the community. Democratizing software creation beyond traditional proprietary platforms.

Examples:
  • Open source core components
  • Community-driven development
  • Shared governance model
5
🔍

Transparent Operations

Users can see and modify how the system works. Complete visibility into platform logic and decision-making processes.

Examples:
  • Inspect all system operations
  • Modify platform behavior
  • Full audit trails and observability

Traditional vs. Fluid Software

🏢Traditional Software

  • Static, predetermined functionality
  • Users interact through pre-built interfaces
  • Limited to what developers anticipated
  • Changes require code deployment cycles
  • Clear separation: developers build, users consume
  • Closed systems (can't see/modify internals)

🌊Fluid Software (Awareness)

  • Natural language to system actions (SQL, file ops, workflows)
  • Users and AI share same tools and permissions
  • Emergent capabilities from component orchestration
  • Memory-enabled agents that learn preferences and patterns
  • Collaborative spaces with files, databases, and workflows
  • Human-in-the-loop approvals for safety

User Intent to Implementation

1

User Expresses Intent

Natural language request: 'Find all invoices from last quarter and summarize spending by category'

2

AI Plans Coordination

Agent decides: search files for invoices → extract data → query database → group by category → summarize

3

Components Execute

File Manager searches MinIO → Data Store runs SQL → Memory stores context → Results aggregate

4

Memory Learns

System remembers: user prefers quarterly summaries, invoice location patterns, spending categories

5

Emergent Automation

Next time: AI proactively suggests 'Want your Q2 spending summary?' before user asks

Real Components, Real Systems

Awareness provides production-ready managers for real-world systems. These aren't simulations or abstractions—they're working with actual files, databases, and services.

📁

File Manager

CAPABILITIES:
  • Multi-backend: Local filesystem, S3-compatible (MinIO), Git repositories
  • CRUD operations with versioning and metadata tracking
  • Text extraction for search (PDF, DOCX, TXT)
  • Hash-based deduplication
  • Natural language search across file contents
EXAMPLE USAGE:

Upload contracts → AI finds clauses → Summarize terms → Store insights in memory

🗄️

Data Store Manager

CAPABILITIES:
  • PostgreSQL adapter with full SQL support
  • Neo4j adapter for graph queries (Cypher)
  • Document store using JSONB columns
  • LLM-powered natural language to SQL translation
  • Schema introspection and metadata discovery
EXAMPLE USAGE:

Connect your database → Ask 'Show me churned customers' → AI generates SQL → Returns results

⚙️

Workflow Engine

CAPABILITIES:
  • Step types: action, condition, loop, parallel, wait, human-approval, AI-decision
  • Variable interpolation and context passing
  • Error handling with retry/skip/goto strategies
  • Cron scheduling for recurring workflows
  • Execution logs and state tracking
EXAMPLE USAGE:

Define multi-step process → AI monitors execution → Trigger on conditions → Human approves critical steps

🔐

OAuth Connection Manager

CAPABILITIES:
  • Standard OAuth 2.0 flows
  • Token storage with encryption
  • Automatic token refresh
  • Provider registry (Google, Slack, GitHub, etc.)
  • AI can use connections for authenticated API calls
EXAMPLE USAGE:

Connect Slack → AI monitors channels → Summarizes threads → Posts updates with your credentials

Foundation vs. Product

Awareness is currently a powerful framework with production-grade backend systems. The managers above are fully functional and tested. What's evolving is the frontend integration—making these capabilities easily accessible through intuitive UIs and seamless AI orchestration.

Think of it as having a professional chef's kitchen (all the tools work perfectly) but still designing the dining experience. The infrastructure is solid; the interface is maturing.

Democratizing Software Development

Awareness aims to make software development accessible to everyone, not just professional developers.

Breaking Down Barriers

  • • No artificial barriers between user and system
  • • Visual builders alongside conversational interfaces
  • • Community ownership vs. proprietary platforms
  • • Full transparency—inspect and modify everything

Empowering Users

  • • Self-hosting removes vendor dependencies
  • • Open source philosophy (MIT/Apache licensed)
  • • Intent-based interaction reduces technical barriers
  • • AI assistance for non-programmers

Dual Interaction Paradigm

💬Mode 1: Conversational Orchestration

Natural language commands that coordinate files, databases, and workflows

"Query our user database for active subscriptions"
"Search my files for contract templates from 2024"
"Create a workflow that notifies me when files are uploaded"

🖱️Mode 2: Direct Management

Traditional UIs for file uploads, database connections, workflow configuration

• Upload files via drag-and-drop
• Configure PostgreSQL connections
• Design workflows with visual builder
• Manage space members and permissions

Key Insight: The AI uses the same File Manager, Data Store adapters, and Workflow Engine that users configure. When you upload a file, the AI can read it. When you connect a database, the AI can query it. When you create a workflow, the AI can trigger it. This creates true parity—same tools, same permissions, same capabilities.

Our Mission

  • 🎯Create a model-agnostic framework that works with any LLM provider
  • 🤝Enable human-AI collaborative development with equal capabilities
  • 🌍Build community-owned platforms, not proprietary SaaS
  • 🔒Maintain privacy and control through self-hosting
  • 💡Make software development accessible to non-programmers