Agentic RAG with Memory || Phidata and Qdrant

[@FarisandFarida]

Awesome 👏. Thank you for this

[@phidata]

Thank you for featuring Phidata, Tarun !!

[@harshasr3705]

My fav sensei back at it making things easy for me!!! Love your content

[@RashomonAI]

Damn, this is soo cool. The inference and quality looks promising. As always loved the flow of video. Thanks for sharing

[@manojkurdekar2400]

Your tutorial are definitely better compared to others I've seen especially with RAG maintenance in general. Thanks

[@changtimwu]

came from qdrant's X repost. The terminal output is so fancy. Maybe make it as a RAG CLI. Just search among personal frequently used knowledge bases.

[@AIwithTarun]

You can use this: agent.cli_app(stream=True) instead of running print_response.

[@AlessioDiasparra-lx4ns]

Hi Tarun I receive ModuleNotFoundError: No module named 'phi, how to fix? Thanks for your content !

[@AIwithTarun]

can you check if you installed `pip install phidata`

[@AlessioDiasparra-lx4ns]

@@AIwithTarun yes i install phidata, the results of pip list : Package Version ----------------- ---------- annotated-types 0.7.0 anyio 4.7.0 certifi 2024.12.14 click 8.1.8 distro 1.9.0 duckduckgo_search 7.1.1 exceptiongroup 1.2.2 gitdb 4.0.11 GitPython 3.1.43 grpcio 1.68.1 grpcio-tools 1.68.1 h11 0.14.0 h2 4.1.0 hpack 4.0.0 httpcore 1.0.7 httpx 0.28.1 hyperframe 6.0.1 idna 3.10 jiter 0.8.2 lxml 5.3.0 markdown-it-py 3.0.0 mdurl 0.1.2 numpy 2.0.2 openai 1.58.1 phidata 2.7.6 pip 24.3.1 portalocker 2.10.1 primp 0.9.2 protobuf 5.29.2 pydantic 2.10.4 pydantic_core 2.27.2 pydantic-settings 2.7.0 Pygments 2.18.0 python-dotenv 1.0.1 PyYAML 6.0.2 qdrant-client 1.12.2 rich 13.9.4 setuptools 58.0.4 shellingham 1.5.4 smmap 5.0.1 sniffio 1.3.1 tomli 2.2.1 tqdm 4.67.1 typer 0.15.1 typing_extensions 4.12.2 urllib3 2.3.0

[@AlessioDiasparra-lx4ns]

@@AIwithTarun yes is installed phidata 2.7.6

[@AIwithTarun]

@@AlessioDiasparra-lx4ns not sure if you are running the virtual environment. Can you join the Discord server: discord.com/invite/hEMqtDXCHA We can take the discussion further over there

[@mihir5846]

Who exactly is making a decision whether to use the knowledge base or tool, openai model? Can a similar thing work with any other model? And how do you make a custom tool and decision making?

[@AIwithTarun]

LLM is doing the decision making part with heavy prompting. It depends on model reasoning capabilities. Gpt4-o, Claude, Grok XAI have good reasoning capabilities. Custom tool logic is possible, it just a function that returns context. Maybe I will make one video on that.

[@mihir5846]

​@@AIwithTarun Thanks for the reply. Looking forward to your future videos.

[@revanthbhuvanagiri9177]

Hey , really excellent work, I have issue while openAI API keys , could you please make video on how to create openAI api keys , it would be really helpful.😁😄

[@AIwithTarun]

- Go to platform.openai.com and login - Click on dashboard and navigate to API keys - Create a new key and save it somewhere. Also to remind you OpenAI API keys are not free, if you have the access then you can use it. If not use Gemini or XAI (grok)

[@anshumanngupta5030]

you could hv explained better with n8n

[@AIwithTarun]

@@anshumanngupta5030 I have not used n8n before. Will check it out once

[@ZainFacts]

Impressive work, I just want ot ask is this work with google gemini model beacuse i have tried it out it always tell me to add OPENAI_API_KEY=

[@AIwithTarun]

In my video I am using OPENAI_API_KEY. In order to use Google Gemini, you need to use GOOGLE_API_KEY. Check out this documentation: docs.phidata.com/reference/model/gemini

[@ZainFacts]

@@AIwithTarun I have tried it but its always give error and force me to add openai api key

[@AIwithTarun]

@ Are you using knowledge base? If yes then that uses embedding model that runs on OpenAI. You need to change the embedding model to Gemini as well

[@AIwithTarun]

vector_db = Qdrant( collection=COLLECTION_NAME, url=QDRANT_URL, api_key=QDRANT_API_KEY, embedder=GeminiEmbedder() ) knowledge_base = WebsiteKnowledgeBase( urls = ["qdrant.tech/documentation/overview/"], max_links = 10, vector_db = vector_db, ) Try this. Import: from phi.embedder.google import GeminiEmbedder