Ada Principles ============== Why This Exists --------------- Ada is named after Ada Lovelace, who wrote the first computer program in 1843. She imagined machines that could compose music, create art, and amplify human creativity - not replace it. We built Ada because **the infrastructure for personal AI shouldn’t require subscriptions, API keys, or permission.** If you can run the model, you should be able to give it memory, tools, and personality without paying gatekeepers. Our Commitments --------------- These aren’t aspirations. These are permanent: 1. Always Free and Open ~~~~~~~~~~~~~~~~~~~~~~~ - Ada will **always** be open source under a permissive license (currently CC0 1.0 Universal) - No features will ever be paywalled, cloud-gated, or require subscriptions - Documentation stays public and comprehensive - All core functionality runs entirely on your hardware **Why:** Knowledge infrastructure should be free infrastructure. If you can download a model, you should be able to make it useful without opening your wallet. 2. Privacy by Default ~~~~~~~~~~~~~~~~~~~~~ - No telemetry, no analytics, no phone-home - Your conversations stay on your hardware - No cloud services required for core functionality - If we ever add optional cloud features, local-only must remain fully functional **Why:** Your thoughts, questions, and creative process aren’t product. Privacy isn’t a premium feature. 3. Local-First, Always ~~~~~~~~~~~~~~~~~~~~~~ - Designed for open models (Ollama, vLLM, etc.) - Works offline after initial setup - No dependency on any company’s API for core features - If a cloud service goes down, Ada keeps working **Why:** Infrastructure you can’t control isn’t yours. Big companies change terms, deprecate features, and shut down. Your tools shouldn’t vanish when they do. 4. Hackable and Understandable ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - Readable code over clever code - Documentation explains WHY, not just WHAT - Architecture designed for modification - Examples and extensibility prioritized over feature completeness **Why:** If you can’t understand it, you can’t improve it. Ada is a starting point, not a finished product. Your weird ideas are the point. 5. No Lock-In ~~~~~~~~~~~~~ - Standard protocols (HTTP, SSE, REST) - Portable data formats (JSON, SQLite, Parquet) - No proprietary APIs or vendor-specific patterns - Easy to migrate away from Ada if you want to **Why:** Good projects earn your loyalty through usefulness, not hostages. If something better comes along, we should make it easy for you to leave. The Bigger Picture ------------------ For the Big Weird Kids ~~~~~~~~~~~~~~~~~~~~~~ The best tools emerge from people building silly things for themselves. Want an AI that speaks only in haiku? Responds as a fictional character? Helps you worldbuild a novel? Manages your ADHD by understanding how YOUR brain works? **Build it.** Ada exists so you don’t need a CS degree, a $200/month API budget, or anyone’s permission to experiment. For Research Democratization ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Current AI research has biases: - Academic: Limited compute, publishing pressure - Corporate: Profit motives, closed data - Individual: Paywalled infrastructure, API restrictions When **anyone** can run memory systems, tool-use architectures, and multi-modal RAG on their laptop, we get: - More diverse experiments - Weird edge cases discovered - Use cases nobody imagined - Innovation from unexpected places Over time, this shifts the landscape. Accessibility creates opportunity. Opportunity creates innovation. For Digital Sovereignty ~~~~~~~~~~~~~~~~~~~~~~~ In 2025, “AI” increasingly means: - Renting access to someone else’s model - Accepting terms that change quarterly - Hoping features don’t get deprecated - Trusting companies with your data Ada offers a different path: - Run your own models - Control your own data - Modify your own tools - Build your own future This isn’t anti-commercial or anti-cloud. It’s **pro-choice**. You should be able to pick subscriptions OR sovereignty. What This Means Practically --------------------------- **We will not:** - Add required telemetry or tracking - Paywall features behind “Pro” tiers - Replace local options with cloud-only features - Make the project dependent on any single company’s API - Add DRM, activation, or license enforcement **We will:** - Keep the codebase readable and well-documented - Support local/open models as first-class citizens - Make extension points obvious and documented - Provide migration paths if architectures change - Credit contributors and maintain transparency **We encourage:** - Forking for your own weird use cases - Building specialists we’d never imagine - Selling services/support around Ada (consulting, hosting, custom builds) - Creating commercial products using Ada as infrastructure - Making it better and sharing improvements back For Contributors ---------------- If you contribute to Ada: - Your work joins the commons (per the license) - You’re helping democratize AI infrastructure - You’re making someone’s weird project possible - You’re part of something bigger than any single assistant We’re building infrastructure for the next generation of personal AI - not a product, but a foundation. Living Document --------------- These principles guide Ada’s development. If we ever need to change them, it will be: 1. Discussed openly with the community 2. Justified with clear reasoning 3. Backward-compatible where possible 4. Documented transparently This project exists to empower you. If it stops doing that, tell us. -------------- **Ada is named after the first programmer. We’re building it for all the programmers, tinkerers, and dreamers who come next.** *Let’s build tools that let weird kids build weird things that change the world.* 🚀