I Built SlothFlow AI: A Local-First AI Assistant That Turns Brain Dumps Into Action Plans 🦥

We all have those days.

You have assignments to finish, projects to build, emails to reply to, interviews to prepare for, and about twenty other things running through your head.

You know what you need to do. The hard part is figuring out where to start.

I wanted to build something that could help with that problem, so for the Hacktoberfest DEV Challenge Week 3, I built SlothFlow AI — a local-first productivity assistant designed to turn overwhelming brain dumps into manageable action plans.

What is SlothFlow AI?

SlothFlow AI follows a simple idea:

Do less. Get more done.

Instead of spending time organizing every thought manually, you can write everything down in plain language and let SlothFlow turn those notes into a draft task plan.

The goal isn't to create another chatbot that tells you to be productive. It's to build a practical tool that helps you take the next step.

What can it do?

1. Turn brain dumps into action plans

Write something like:

I need to finish my project report by tomorrow, prepare for a backend interview, reply to my mentor, and organize my downloads folder.

SlothFlow converts your notes into individual tasks, suggests priorities, and provides estimated completion times.

You can review the results and choose which tasks to save.

1. Manage tasks in one place

Once saved, tasks appear in your workspace, where you can:

Track pending and completed tasks.

View estimated time remaining.

Mark tasks as complete.

Delete tasks you no longer need.

Add new tasks manually.

1. Start fresh with Daily Reset

The Daily Reset view summarizes your unfinished work and estimated time requirements.

Rather than treating unfinished tasks as failures, the idea is to make it easier to review them and choose a realistic next step.

1. Run AI locally

One of the decisions I made while building SlothFlow was to support local AI inference through Ollama.

When a compatible local model is available, SlothFlow uses it to generate a more contextual task plan. If Ollama isn't available, a built-in rules-based fallback can still produce a basic checklist.

This means you can try the application without an API key or a paid AI service.

How I built it

The project uses a lightweight stack:

Python — application logic

FastAPI — REST API and backend

Ollama — optional local LLM inference

SQLite — local task persistence

HTML, CSS and JavaScript — responsive frontend

Pytest — automated testing

The backend exposes endpoints for generating plans, creating and retrieving tasks, updating task status, deleting tasks, and generating the Daily Reset summary.

I also added input validation and made the AI output pass through a validation step before the application uses it.

Why local-first?

Personal notes can contain unfinished ideas, work details, and other information people might not want to send to an external AI service.

SlothFlow is designed to keep task storage on the machine running the application and to use a locally configured Ollama server for AI inference.

There is no external AI API requirement.

Of course, local-first doesn't automatically mean every deployment is secure. The current project is a local prototype, and a public deployment would need additional security measures.

Challenges along the way

One of the interesting parts of this project was making the application useful even when the AI model wasn't available.

Rather than making the entire application depend on a running LLM, I implemented a fallback rules engine. This keeps the core workflow accessible while allowing users to enable local AI when they want more contextual results.

I also built the task management flow around review and confirmation: generated tasks remain drafts until the user chooses to save them.

That distinction matters. AI-generated suggestions shouldn't silently become someone's to-do list.

What's next?

SlothFlow is an MVP, and there is plenty of room to improve it.

Some features I'd like to explore next include:

Smarter task decomposition and dependency detection.

Better scheduling based on available time.

Markdown export and import.

More robust testing with different local models.

Optional reminders and recurring tasks.

Improved evaluation of task quality and time estimates.

I'd also like to measure how much time the tool actually saves compared with organizing the same notes manually, rather than assuming that automation is automatically better.

Try it and contribute

I'm building SlothFlow as part of the Hacktoberfest DEV Challenge, and I'd love feedback from developers working on local AI, productivity tools, and open-source applications.

GitHub repository:https://github.com/Samearth17/slothflow.ai
Youtube Demo Video:https://youtu.be/CaHC42t-EOk

If you try it, I'd love to hear what works, what doesn't, and which repetitive task you'd want to automate next.

If you'd like to contribute, ideas around better task extraction, accessibility, testing, and local model support would be especially welcome.

Final thoughts

I think the most useful AI applications aren't necessarily the ones that do the most complicated things. Sometimes, they're the ones that remove a small but persistent source of friction from everyday life.

That's what I wanted to explore with SlothFlow AI.

Less time organizing the work.

More time actually doing it.

Thanks for reading! 🦥