Our target for starting strong in 2026 #375
Replies: 3 comments 1 reply
|
Can I contribute to your project using Rust? |
|
It's not an issue for us to use Rust.It's just that we need a community for supporting that. And today we barely have it running 100% for Python/NextJS yet. So we might need to work on our contributor network first before taking on new tech. If you can write Python/Next, you'd be very welcome :) |
|
Closing this as a completed historical announcement. It remains an important snapshot of where Open Notebook started 2026, but the project direction and contribution process have continued to evolve. For the current product direction, see VISION.md. For the current community workflow and where to participate, see Welcome to Open Notebook Discussions. Thank you to everyone who helped move these goals forward. |
Uh oh!
There was an error while loading. Please reload this page.
Happy New Year, Open Notebook community!
Like many of you, I've been thinking about resolutions for 2026. And I want to share one that directly affects this project:
This year, I'm committing to dedicating significantly more time to Open Notebook — not just coding, but building the infrastructure for more people to participate.
Looking Back at 2025
What a year. We hit 18,000 stars on GitHub. The community grew beyond anything I imagined when I started this project. People are using Open Notebook for research, learning, privacy-focused AI workflows, and things I never anticipated.
But I'll be honest: things got a bit messy.
The issue tracker became overwhelming. Priorities weren't clear. Contributors wanted to help but didn't know where to start. I was reactive instead of proactive. The project grew faster than my ability to organize it. And the day-to-day responsibilities keeped me from spending that time that I wanted here.
That changes now.
The Bigger Picture
Two things are pushing me to accelerate:
AI Evolution isn't standing still
Big Techs are shipping features at a rapid pace. New interaction modes, better audio, more integrations. We can't afford to fall behind. Open Notebook exists because people want an alternative they control — but "alternative" doesn't mean "inferior." We need to match and exceed the experience, while keeping our core values: privacy, self-hosting, openness.
AI-assisted development changes everything
More developers now use AI coding assistants — Claude Code, Cursor, Copilot, Windsurf. This is an opportunity. If our codebase is well-organized, well-documented, and has clear contribution paths, these tools can help contributors be 10x more productive.
I'm investing heavily in developer ergonomics:
The goal: a developer with an AI assistant should be able to pick up an issue and contribute meaningfully within hours, not days.
What should we be working On
I spent the first day of 2026 mapping out every challenge facing Open Notebook. Here's what's on our radar:
1. RAG & Context Mechanism
Problem: We inject full source content into conversations. This overloads models and produces subpar responses — even though we already have content chunked and embedded.
Direction: Build proper RAG using semantic search to select relevant chunks.
2. Documentation & Onboarding
Problem: Setup is harder than it should be. Docker configs are confusing.
Direction: Rewrite docs for "get started in 5 minutes." We're growing despite this friction — imagine what happens when we remove it.
3. Ollama & Local LLMs
Problem: Integration has issues — some config-related, some deeper (agents not understanding context).
Direction: Deep investigation into message formatting. This matters too much to privacy-focused users.
4. SurrealDB
Problem: Transaction conflicts under load, production readiness questions.
Our decision: We're staying with SurrealDB. It gives us document storage, graph relationships, vector embeddings, and background jobs — all in one. The alternative (Postgres + Redis + Celery + vector DB) doesn't align with "easy self-hosting."
Full decision documented in #372. If you have SurrealDB production experience, we'd love your input.
5. Sources & Content
Problem: Performance degrades with many sources. UI has issues.
Direction: This is the heart of the product. Prioritizing fixes here.
6. Podcast Feature
Problem: Multiple bugs, inconsistent experience.
Direction: Root cause analysis, then redesign if needed.
7. Internationalization
Problem: Growing demand for Chinese (and other languages).
Direction: Setting up i18n infrastructure. Community translations welcome.
8. Offline / Air-Gapped Support
Problem: tiktoken downloads tokenizers on first run.
Direction: Pre-cache in Docker image. See #374.
9. Deployment Options
Problem: People want Kubernetes, EasyPanel, one-click deploys.
Direction: Helm chart PR in review (#363). EasyPanel template ready for pickup.
10. AI Providers
Problem: Requests for new providers.
Direction: On hold until we decide on LLM abstraction (evaluating LiteLLM migration).
11. Multi-User Support
Problem: Teams want shared access.
Direction: Not now. Individual users first. Foundation needs to be solid.
How You Can Participate
If you code
We've created umbrella issues with clear specs:
More coming for i18n, UI fixes, and deployments.
Each umbrella has context, decisions made, sub-tasks with difficulty, and getting started instructions. Claim a task by commenting.
If you don't code
If you have expertise
We're looking for people with experience in:
A Personal Note
Open Notebook started as a personal project — a tool I built because I wanted something that didn't exist. Watching it grow to serve thousands of people has been humbling.
But growth without organization is chaos. This year, I'm building the infrastructure — not just technical, but human — so this community can thrive.
Thank you for being part of this. Let's make 2026 the year Open Notebook becomes not just popular, but excellent.
Happy New Year,
Luis
P.S. — If you've read this far, drop a comment below. Even just a "hello" from your part of the world. I'd love to know where Open Notebook has reached.
All reactions