Lateral Thinking with Proven Technology
âInnovation is not about using the newest technology, but about using technology in new ways.â â Adapted from Nintendoâs philosophy
Our Approach
Section titled âOur ApproachâAt Jan, we donât chase the bleeding edge for its own sake. We take proven, battle-tested technologies and combine them in innovative ways to solve real problems. This philosophy shapes everything we build.
Why Proven Technology Wins
Section titled âWhy Proven Technology WinsâStability Over Novelty
Section titled âStability Over Noveltyâ- GGUF Format: We didnât invent a new model format. We adopted the community standard that already works across platforms
- llama.cpp: Years of optimization by brilliant mindsâwhy reinvent when we can build upon?
- Tauri + Web Tech: Proven UI stack that works everywhere, letting us focus on AI innovation
- SQLite: The worldâs most deployed database for local dataâperfect for privacy-first architecture
Accessibility Through Maturity
Section titled âAccessibility Through MaturityâWhen we use proven technology:
- Hardware requirements are well understood
- Optimization paths are clear
- Community knowledge exists
- Documentation is comprehensive
- Edge cases are known
This means Jan works on more devices, for more people, with fewer surprises.
Lateral Innovation Examples
Section titled âLateral Innovation ExamplesâLocal AI That Actually Works
Section titled âLocal AI That Actually WorksâTraditional approach: Build massive cloud infrastructure, require constant internet Our approach: Use proven quantization techniques to run powerful models on consumer hardware
Privacy Without Compromise
Section titled âPrivacy Without CompromiseâTraditional approach: Complex encryption and privacy theater Our approach: Simply donât collect data. Use local SQLite. Problem solved.
Universal Model Support
Section titled âUniversal Model SupportâTraditional approach: Proprietary model formats and exclusive partnerships Our approach: Support the open GGUF standard that the community already uses
Cross-Platform Without Complexity
Section titled âCross-Platform Without ComplexityâTraditional approach: Native apps for each platform, massive development overhead Our approach: One codebase using Electron, works everywhere, ships faster
The Power of Constraints
Section titled âThe Power of ConstraintsâWorking with proven technology creates healthy constraints:
Resource Efficiency
Section titled âResource Efficiencyâ- Canât throw infinite compute at problems
- Must optimize for consumer hardware
- Forces elegant solutions over brute force
- Makes us accessible globally, not just in rich markets
Clear Boundaries
Section titled âClear Boundariesâ- Known limitations guide design decisions
- Realistic about whatâs possible today
- Focus on solving real problems now
- Build stepping stones to the future
Community Alignment
Section titled âCommunity Alignmentâ- Use formats the community knows
- Build on protocols people trust
- Contribute improvements back upstream
- Stand on the shoulders of giants
Innovation Through Integration
Section titled âInnovation Through IntegrationâOur innovation comes from how we combine proven pieces:
llama.cpp (proven inference) + GGUF (standard format) + Electron (cross-platform UI) + SQLite (local data) + MCP (extensibility) = Jan (accessible AI for everyone)
Each component is battle-tested. The magic is in the combination.
Real Problems, Real Solutions
Section titled âReal Problems, Real SolutionsâThis philosophy keeps us grounded:
What We Build
Section titled âWhat We Buildâ- AI that runs on the laptop you already own
- Privacy that doesnât require a PhD to understand
- Tools that work offline in rural areas
- Features users actually need today
What We Donât Build
Section titled âWhat We Donât Buildâ- Exotic architectures that need exotic hardware
- Complex systems that break in simple ways
- Features that demo well but fail in practice
- Solutions looking for problems
The Compound Effect
Section titled âThe Compound EffectâUsing proven technology creates compounding benefits:
- Faster Development: Less time debugging novel approaches
- Better Reliability: Years of battle-testing by others
- Easier Adoption: Users already understand the patterns
- Stronger Ecosystem: Can leverage existing tools and knowledge
- Lower Costs: Commodity hardware and proven optimizations
Learning from History
Section titled âLearning from HistoryâGreat innovations often use âoldâ technology in new ways:
- The iPhone used existing touchscreen tech
- Tesla started with laptop batteries
- SpaceX used proven rocket designs more efficiently
- The web succeeded using simple protocols
Jan follows this tradition: proven AI inference, standard formats, and simple principlesâcombined in a way that makes AI accessible to everyone.
Building for the Long Term
Section titled âBuilding for the Long TermâProven technology has staying power:
- Standards that last decades
- Protocols that survive company failures
- Formats that outlive their creators
- Communities that maintain forever
By building on these foundations, Jan can focus on what matters: making AI useful, private, and accessible to everyone.
Our Commitment
Section titled âOur CommitmentâWe commit to:
- Practical over Perfect: Ship what works today
- Simple over Sophisticated: Elegant solutions to real problems
- Proven over Promising: Build on solid foundations
- Accessible over Advanced: Reach everyone, not just enthusiasts
The Future is Already Here
Section titled âThe Future is Already HereâThe technologies needed for private, local AI already exist. Theyâre proven, optimized, and ready. Our job isnât to invent new technologyâitâs to combine what exists in ways that serve humanity.
Thatâs lateral thinking. Thatâs how we build the future: not by reaching for tomorrowâs technology, but by using todayâs technology in tomorrowâs ways.
âThe best technology is invisible. It just works, everywhere, for everyone.â