The Sundarban
Printed Feb 22, 2026, 5:00 PM EST
Raghav Sethi began his tech writing lope in 2022, contributing to his faculty’s initiate-provide community weblog. Later that year, he joined MakeUseOf, and since then has written extensively about Apple, Android, and AI. His work ranges from hands-on experiments to opinion objects that explore the greater narrate within the back of rising tech traits.
Alongside his work at MUO, you can also obtain Raghav’s articles at XDA Builders, the place he mainly makes a speciality of Linux and the area of initiate-provide software.
Commence air of writing, Raghav enjoys engaged on coding projects, playing the guitar, and living existence on the brink by installing the latest beta software on his daily units.
Whereas NotebookLM is an amazing research instrument, and there are some great prompts that can make it even extra necessary, I’ve always felt to find it irresistible was a bit too mighty for my wants, even supposing that sounds love an unpopular opinion. I wanted one thing that sits within the middle, now not as minimal as the ChatGPT web location, nonetheless now not as heavy or structured as NotebookLM.
That’s when I started exploring Khoj, (which translates to “discovery” in English). After spending some time with it, I actually really feel love I’ve came across that most attention-grabbing heart floor I was searching for.
Related
Khoj AI
Khoj AI is a web research assistant that helps you rapidly obtain and summarize information from the procure. It combines necessary search and context-aware responses to make research easier and extra ambiance friendly.
NotebookLM is brilliant, nonetheless I love this extra
Initially, there are two ways to make use of Khoj AI. You can either use the procure location, the place the free tier has access to basic models love Gemini Flash 3, otherwise you can self-host and bid your gain model. To obtain started, I may probably recommend that you strive out the procure location first. There may be now not any point in surroundings everything up locally whereas you do not halt up liking it.
When you initiate the procure location, it looks very similar to Claude, so that you’ll really feel factual at dwelling. There’s a chat field along with various features for writing code, analyzing documents, and extra. Basically, it’s ready to handle most “research-heavy” tasks factual away.
Above the chat field, you are going to also gape extra than one agents. The default is Khoj, nonetheless you can swap over to one thing love technical lead, a teacher, and even a legal expert. Apt pick which one you want and it is ready to head!
Keep in mind that LLMs can hallucinate, no matter how capable the model is. Always double-examine important information, especially when it comes to areas love legal or medical advice.
One among the really frigid things Khoj affords is slash commands. Apt fashion “/” and you’ll gape a listing of commands that you can add to your advised.
For example, typing “/notes” forces the LLM to drag information handiest out of your gain documents. Using “/code” launches the built-in Python interpreter, which can even generate graphs and varied figures straight at some point of the app using Matplotlib. The correct way to characterize this app may probably be if NotebookLM and Claude had a baby.
You can even produce your gain agents and automations
Carry out even extra than research
I mentioned earlier that Khoj AI entails built-in agents that can take on varied roles, love acting as a coach or a legal expert. However what makes it even extra necessary is that you can create your gain agents tailored to your suppose workflows.
Within the Agents tab, pick Create Agent, and you’ll be brought about to enter a few basic details. After that, you can add information to its data base, settle the model, pickle the enter and output modes, and you’re all pickle.
However clearly, as I mentioned earlier, the accuracy of your agent is largely going to count on the model you selected. If one thing goes immoral, it is extra the model’s fault than Khoj AI’s.
Apt love agents, you can also create your gain automations that hasten tasks automatically at a specified time (duh.) For example, you may obtain a full summary of the stock market every day at 9:00 AM, or pickle up an agentic-esque RSS feed that fetches the latest tales at a scheduled time.
There are masses of pre-built automations available, nonetheless you can also create your gain. Apt tear to the Automations tab and pick Create Automation. From there, settle the days and occasions it should hasten, and specify what information you want the LLM to generate.
That’s it! You’ll obtain the information at the same time every day straight to your email, with out writing a single line of code.
You gain your complete stack whereas you happen to self-host Khoj AI
Maintain, don’t lease
I obtain it whereas you do not want to count on the cloud, and the accurate thing is you can self-host Khoj via Docker. It can even call your LLMs via Ollama, so that’s a monumental hassle taken care of as correctly. To obtain started with operating it to your gain hardware, be certain that you have Docker installed to your system. As soon as that is achieved, simply hasten the following command for your terminal:
mkdir ~/.khoj && cd ~/.khoj wget https://raw.githubusercontent.com/khoj-ai/khoj/main/docker-gain.yml
Now, you are going to have to configure some things within the docker-gain.yml file. You can originate so by coming into this command:
nano docker-gain.yml
Commence the file and survey for KHOJ_ADMIN_EMAIL and KHOJ_ADMIN_PASSWORD. Replace them with the email and password you want to make use of for your admin account.
When you plan to make use of a third-party LLM provider, you can also add your API keys for OpenAI, Gemini, or Anthropic of their respective fields within the same file. After that, you may be graceful mighty achieved! Apt be certain that you may be calm within the same itemizing for your terminal and enter the following command:
docker-gain up
By default,
