Meet Potion: Your Smart Note-Taking Companion

Hello Everyone,

In this article, we are going to know how i built my latest project Potion with MindsDB, And how it made it super easy to build the advanced AI integrated project from both user and developer side. (Yes! it is)

First of all, What is MindsDB?

Before going deeper into other details, let's first understand what MindsDB is and how it made it super easy to build the workflow and execute it. MindsDB is an AI Data Solution Platform that makes it easy to connect, unify, and respond (as stated on their site). Simply put, we can understand MindsDB as an abstraction (Hidden Layer) that handles the connection of various data sources and automatically creates pipelines from that data, so you don't need to worry about AI integration development. MindsDB made it easy, like plug and play. Now, you can focus more on business logic rather than other unnecessary stuff like building a chatbot that answers user queries (sadly, but it's the truth!).

Core Principle of MindsDB

seamless integration, organization, and utilization of data (same thing we understand above :))

But how MindsDB able to bring all these features...What's the secret?

SQL is the secret (Yes!). MindsDB is using SQL-based syntax as its core interface for AI and machine learning operations. By enhancing standard SQL with AI-specific extensions, MindsDB makes artificial intelligence accessible through a familiar database query language. This approach democratizes AI, allowing developers and data professionals to use their existing SQL skills for machine learning tasks, thus removing traditional barriers to AI adoption.

This approach not only simplifies the integration process but also ensures that AI capabilities are seamlessly incorporated into existing workflows with minimal effort. As we all know, SQL has historically transformed database querying, saving billions for companies worldwide. Now, MindsDB is harnessing that power to drive their advanced AI innovation, making complex integrations more efficient and accessible.

During the development of Potion, I realized why this is crucial for developers and how it greatly impacts the development cycle, making a developer's work much easier and more efficient.

What is Potion, through which I learned all of this?

Potion is an Intelligent Note-Taking Assistant that stores your daily notes or moments in your local storage (yes! no database, 100% privacy) and your local MindsDB instances or image if you are using Docker :). But this is a simple note application—what's intelligent about it? Wait, Potion not only stores your notes, it can search your notes (not just a keyword matching search), it's an AI Semantic Search. You have an AI Agent running only for you and with your data locally (Yes! No more clouds and data privacy issues). The story doesn't end there; you also get an AI summarization feature that can summarize your notes into a much more concise form (Yes! No more juggling). Sounds cool, doesn't it?

All of these features are too complex, and generally, it takes me months to implement all of this in a single application. But can you believe, I shipped all of these features in a week (yes! working week). And it stretched to a week because I got some logic errors; otherwise, you can ship all of these features in your applications in less than a week. (yes! it is true, no bluffs). Thanks to MindsDB for bringing such a platform that can help you ship industry-level features within a span of a day.

Some of the key features that i like in MindsDB?

MindsDB and their platform are loaded with so many cool features, you can check out later by installing it locally, but the features that I like the most are:

1. Knowledge Base: Kb is one of the finest and best features of MindsDB. Using Knowledge Base, you can transform your raw data into vector embeddings that can help you build various applications based on it, like RAG, Text Summarization, and much more. And for all of this, you do not need complex implementations, integrations of Vector DB, Embedding Models, or other stuff that can make you mad. But with MindsDB, you just need these code snippets and that’s it! Now add your data and build your application. (Yes! it is that simple)

   CREATE KNOWLEDGE_BASE my_kb
   USING
       embedding_model = {
           "provider": "openai",
           "model_name" : "text-embedding-3-large",
           "api_key": "sk-abc123"
       },
       reranking_model = {
           "provider": "openai",
           "model_name": "gpt-4o",
           "api_key": "sk-abc123"
       },
       metadata_columns = ['product'],
       content_columns = ['notes'],
       id_column = 'order_id';
   ```

2. **Agent**: Earlier, when we needed to add an AI Agent or build an AI Agent, we first had to create a workflow for it and implement it. If we wanted it to be based on our data, we either needed to fine-tune that LLM model (which is too costly) or build an efficient pipeline that could handle such specific tasks (not an easy job). Both of these options are too hectic and costly, plus it is not guaranteed that your Agent will work perfectly unless you do extensive testing (again, a hectic and costly process).

   But MindsDB brings a solution to eliminate this problem, which is the MindsDB Agent. MindsDB’s Agents are industry-level conversational agents that can answer questions using your data with higher accuracy. To use a MindsDB Agent, you do not need high-level functions or implementations. With a simple query, you can deploy an industry-level Agent.


```sql
CREATE AGENT my_agent
USING
   model = 'gemini-2.0-flash',
   google_api_key = 'xyz123',
   include_knowledge_bases= ['mindsdb.sales_kb', 'mindsdb.orders_kb'],
   include_tables=['postgres_conn.customers', 'mysql_conn.products'],
   prompt_template='
       mindsdb.sales_kb stores sales analytics data
       mindsdb.orders_kb stores order data
       postgres_conn.customers stores customers data
       mysql_conn.products stores products data
   ';

There are many more exciting features available in MindsDB, like Models, AI Tables, Views, Jobs, and much more. If you are thinking of adding AI power to your application, then do check out MindsDB; you will be glad you did. Yes!

Now, let’s come to potion again..

Potion is a fully open-source web application, and all the code is available in the GitHub repo. Do check it out if you want to explore MindsDB. But keep in mind, you should have MindsDB installed locally first to use all the core AI features of Potion. For details, you can check out these guides.

Here’s the Repo: Potion’s GitHub Repo

Please give it a star ⭐ if you like the project. It really boosts my energy to create more applications like this for all of you.

If you don’t have time to install and try everything, no worries! We also have a Demo Video that shows you a glimpse of all the awesome MindsDB integrations in Potion. However, not everything can be covered in a demo. Therefore, checking out the codebase will surely give you an extra edge. (And we are developers, we can read codebases.)

Here’s the Demo: Potion’s Demo Video

I don’t need to say this, but still, if you like the demo, please like the video. If you don’t like the video, you can also dislike it and tell me in the comments what I can improve in my demos.

Thank You!

If you reached this point, I want to thank you. Thanks for reading this blog. I hope you found something new in this article that can help you in some way. If you have used MindsDB, please let me know about your experience. And if not, try it first and then tell me how much you like it. ;)

Good Bye!!