Chapters: 
AI

You can find various examples of Chainlit apps here that leverage tools and services such as OpenAI, Anthropiс, LangChain, LlamaIndex, ChromaDB, Pinecone and more.

📃 License

Chainlit is open-source and licensed under the Apache 2.0 license. 

[CML] tux@camelot …/ai-agents-crash-course/multi_agent_chatbot $ sh run_local.sh
run_local.sh: line 2: chainlit: command not found

[CML] tux@camelot …/ai-agents-crash-course/multi_agent_chatbot $ which python
/usr/bin/which: no python in (/home/tux/.local/bin:/home/tux/bin:/usr/local/bin:/usr/bin)
[CML] tux@camelot …/ai-agents-crash-course/multi_agent_chatbot $ which python3
/usr/bin/python3

SET UP venv LOCALLY
=====================
[CML] tux@camelot …/ai-agents-crash-course/multi_agent_chatbot $ python3 -m venv venv
[CML] tux@camelot …/ai-agents-crash-course/multi_agent_chatbot $ ls -lsa
total 32
0 drwxr-xr-x.  4 tux tux  167 Mar 12 11:45 .
4 drwxr-xr-x. 18 tux tux 4096 Mar 11 22:09 ..
4 -rw-r--r--.  1 tux tux 2344 Mar 11 21:32 agentic_chatbot.py
0 drwxr-xr-x.  3 tux tux   45 Mar 11 21:32 .chainlit
4 -rw-r--r--.  1 tux tux  761 Mar 11 21:32 chainlit.md
0 -rw-r--r--.  1 tux tux    0 Mar 11 21:32 __init__.py
12 -rw-r--r--.  1 tux tux 9796 Mar 11 21:32 nutrition_agent.py
4 -rwxr-xr-x.  1 tux tux   76 Mar 11 21:32 run_chatbot.sh
4 -rwxr-xr-x.  1 tux tux   59 Mar 12 11:43 run_local.sh
0 drwxr-xr-x.  5 tux tux   92 Mar 12 11:45 venv

(venv)
=======
[CML] tux@camelot …/ai-agents-crash-course/multi_agent_chatbot $ source venv/bin/activate


(venv) [CML] tux@camelot …/ai-agents-crash-course/multi_agent_chatbot $ which python
~/camelot/ai-agents-crash-course/multi_agent_chatbot/venv/bin/python

requirements.txt exists 

pip install -r ../requirements.txt

 

Load the data

Status check

  • The local venv Python environment works.
  • Chainlit works.
  • The source dataset exists: data/calories.csv.
  • The Chroma database exists but is empty.

So the one task left is simply:

load the CSV into the Chroma collection called nutrition_db.

Once that exists, your chatbot’s lookup tool will work.

What you need to do (conceptually)

  1. Read data/calories.csv.
  2. For each row, extract:
    • food name
    • category
    • calories per 100g
  3. Store those as metadata in Chroma.
  4. Save them under a collection called nutrition_db.

Then your agent can query it.

The only command you’ll run after creating the loader script

From the multi_agent_chatbot directory: 

python build_nutrition_db.py

That will populate ../chroma.

How you'll know it worked

When you check the collections, you should see: 

nutrition_db

After that you can run your chatbot normally.

And just to say it explicitly: your instinct earlier—

“Where does the data come from?”

—is exactly the right question. That’s the difference between running an AI demo and building a system you control.

If you'd like, next we can also make your dataset pipeline cleaner so your repo structure becomes something like: 

data/
vector_db/
build_scripts/
agents/

which is the pattern used in production RAG systems.