docs: add local quickstart #218
Conversation
6eba3e1 to
67ee181
Compare
|
|
||
| --- | ||
|
|
||
| ## Step 1: Set up a database |
There was a problem hiding this comment.
Should their be a "before you begin" step? Are we assuming the user has any tools installed or running a specific environment?
There was a problem hiding this comment.
Right now, we're not assuming anything about the user environment, except that they have python installed.
We're going over the whole process from scratch.
| [Install postgres and configure a | ||
| database](https://neon.tech/postgresql/postgresql-getting-started) for your | ||
| system. | ||
|
|
||
| This process creates a database `postgres` with superuser `postgres`. |
There was a problem hiding this comment.
I would like to see if we can capture this set up ourselves, without requiring the user to go to a new page. I think using docker might also be useful in streamlining?
There was a problem hiding this comment.
In that case, can we assume basic familiarity with docker and have docker installation as a prerequisite?
Do you think the friction of docker installation is worth ignoring?
| 1. Create a LangGraph [ReAct | ||
| agent](https://langchain-ai.github.io/langgraph/reference/prebuilt/#langgraph.prebuilt.chat_agent_executor.create_react_agent). | ||
|
|
||
| ```python | ||
| from langgraph.prebuilt import create_react_agent | ||
| from langchain_google_vertexai import ChatVertexAI | ||
| from langgraph.checkpoint.memory import MemorySaver | ||
|
|
||
| model = ChatVertexAI(model_name="gemini-pro", project="my-project") # Change the GCP project here | ||
| agent = create_react_agent(model, tools, checkpointer=MemorySaver()) | ||
| ``` | ||
|
|
||
| 1. Define the initial prompt and user queries. | ||
|
|
||
| ```python | ||
| prompt = """ | ||
| You're a helpful hotel assistant. You handle hotel searching, booking and | ||
| cancellations. When the user searches for a hotel, mention it's name, id, | ||
| location and price tier. Always mention hotel ids while performing any | ||
| searches. This is very important for any operations. For any bookings or | ||
| cancellations, please provide the appropriate confirmation. | ||
| Don't ask for confirmations from the user. | ||
| """ | ||
|
|
||
| queries = [ | ||
| "Find hotels in Basel with Basel in it's name.", | ||
| "Can you book the Hilton Basel for me?", | ||
| "Oh wait, this is too expensive. Please cancel it and book the Hyatt Regency instead.", | ||
| "My check in dates would be from April 10, 2024 to April 19, 2024.", | ||
| ] | ||
| ``` | ||
|
|
||
| 1. Run the queries and observe the output! | ||
|
|
||
| ```python | ||
| config = {"configurable": {"thread_id": "thread-1"}} | ||
|
|
||
| for query in queries: | ||
| inputs = {"messages": [("user", prompt + query)]} | ||
| response = await agent.ainvoke(inputs, stream_mode="values", config=config) | ||
| print(response["messages"][-1].content) | ||
| ``` | ||
|
|
||
| To verify the agent's actions, you can examine the hotels table. You should | ||
| observe that the `booked` column for the `Hyatt Regency Basel` has changed | ||
| from `0` to `1`, indicating that the hotel has been successfully booked. | ||
| Additionally, the `checkin_date` and `checkout_date` have been updated to | ||
| `2024-04-10` and `2024-04-19` from `2024-04-02` and `2024-04-20` | ||
| respectively. |
There was a problem hiding this comment.
Would it be better to combine this into a single file users can copy and hack on locally?
There was a problem hiding this comment.
This structure could help users understand and customize the steps involved.
However, we could upload the code and YAML files into the github repo and add a link for users to access them.
We could add a note at the beginning, like:
The code and configuration files used in this quickstart are available [here](link_to_files).
What do you think?
| ```python | ||
| prompt = """ | ||
| You're a helpful hotel assistant. You handle hotel searching, booking and | ||
| cancellations. When the user searches for a hotel, mention it's name, id, |
There was a problem hiding this comment.
| cancellations. When the user searches for a hotel, mention it's name, id, | |
| cancellations. When the user searches for a hotel, mention its name, id, |
| """ | ||
|
|
||
| queries = [ | ||
| "Find hotels in Basel with Basel in it's name.", |
There was a problem hiding this comment.
| "Find hotels in Basel with Basel in it's name.", | |
| "Find hotels in Basel with Basel in its name.", |
| "Find hotels in Basel with Basel in it's name.", | ||
| "Can you book the Hilton Basel for me?", | ||
| "Oh wait, this is too expensive. Please cancel it and book the Hyatt Regency instead.", | ||
| "My check in dates would be from April 10, 2024 to April 19, 2024.", |
There was a problem hiding this comment.
| "My check in dates would be from April 10, 2024 to April 19, 2024.", | |
| "My check-in dates would be from April 10, 2024 to April 19, 2024.", |
|
Superseded by #232 |
This is a continuation of #218, which add a local quickstart for running Toolbox with Python and LangGraph. --------- Co-authored-by: Twisha Bansal <twishabansal@google.com> Co-authored-by: Anubhav Dhawan <anubhav756@gmail.com>
|
🧨 Preview deployments removed. |
This quickstart helps users to onboard to toolbox quickly.