Airline AI Agent with Langflow and DataStax Astra DB

Follow the instructions in YouTube:

• Database setup
• Creating Langflow

Database Setup

You should have a DataStax Astra account and then create a database.

To create an account, you can use this link: https://bit.ly/ds-langflow

After creating an account, you can create a database.

Loading the flight tickets data

Access the database and create a collection flight_tickets. After that, you can load the flight tickets data into the collection using the file data/tickets.json.

NOTE: If you want to generate a new file, use this prompt along with your favorite LLM:

generate a sample json file with flight tickets considering the following attributes:

_id 
passengerId
passengerName
flightNumber
departureAirport
arrivalAirport
departureDateTime
arrivalDateTime
seatNumber
ticketClass
ticketPrice
bookingDate
bookingReference

Loading the invoice data

On the Astra DB Dashboard, access the CQL Console to create a regular Cassandra table. Run the following CQL commands:

CREATE TABLE default_keyspace.invoices (
    customer_id uuid,
    flight_id uuid,
    invoice_id uuid,
    installment_due_dates map<int, timestamp>,
    installment_plan map<int, decimal>,
    issue_date timestamp,
    payment_method text,
    status text,
    total_amount decimal,
    PRIMARY KEY (customer_id, flight_id, invoice_id)
) WITH CLUSTERING ORDER BY (flight_id ASC, invoice_id ASC);

After that, you can load the invoice data into the table running the commands from the file data/invoice.sql on the CQL Console.

Creating the Langflow Agent Flow

You can follow the instruction in the video. But here is the JSON file you can import into Langflow to create the flow:

Airline Agent.json

Langsmith

You can use the Langsmith to track the flow. Here is the link to the Langsmith dashboard: https://smith.langchain.com/

To enable the Langsmith, you can set the environment variable LANGCHAIN_TRACING_V2=true and LANGCHAIN_API_KEY=lsv2_....

Contact

If you have any questions, please contact me at:

• Email: samuel.matioli@datastax.com
• LinkedIn: @samuelmatioli
• Website: langflow.com
• DataStax: datastax.com