This is a submission for the Agent.ai Challenge: Full-Stack Agent (See Details)
Hey there everyone!
I am excited to share my submission for the Agent.Ai challenge under the full-stack agent category. Please leave a like and a comment on how I can improve the agent further.
Background
We all love travelling and exploring new places. However, planning for a trip is not so easy - with so many variables and effort to ensure we find the cheapest flights, hotels, etc and, most importantly ensuring we don't miss any good sightseeing places in that place.
Solution
What if we had an agent that could suggest the cheapest flights available for our travel dates, good sightseeing places, local phrases to ease our journey?
What I Built
Enter Wander Guide - Your AI powered travel guide for all your travel needs.
Wander Guide Technology Stack
- Utilises Google Flights API for fetching latest flight data
- Agent.AI Agent Builder tool for setting up the agent workflow
Step-by-Step Walkthrough of development
- Flights Data
In order to fetch the latest data on available flights and price options, I utilised the Google Flights API. This however, was not directly accessible so I developed my own web-service using FastAPI and hosted it on Render. The free tier has a built-in inactivity shut-down which slows down the response time, hence I also created a Cronjob to ping my web-service every 10mins to ensure I get a 100% uptime of my Flights API at zero cost 😉. - Building the Agent
The other parts were done directly using Agent.AI's Agent builder tool. With a mix of user input steps, to processing the answers using AI and then finally presenting the response back to the user.
Demo
The agent can be invoked from this link
Upcoming improvements
[ ] Adding Hotel API
[ ] Building a complete end-to-end trip (including non-flight journey)
[ ] Adding ability to book tickets from the agent itself
[ ] Ability to invoke the agent from whatsapp / telegram
[ ] Email summary of response
Agent.ai Experience
Agent.AI Agent builder tool is extremely user-friendly and intuitive.
I was able to develop the Agent in 2-3 days including setting up a web-service for the flights API. Here are few points that I would like to highlight.
- Handling data-input from users
One of the most challenging tasks, in my opinion, while creating traditional chatbots was handling user responses. For example, in this agent, the date input from the user can be in any format - DD/MM/YY or MMM/DD/YYYY, etc. Agent.AI's ability to use GenAI models makes this seem like a breeze - with a little bit of prompt engineering, one can practically make all data processing steps extremely easy. - Integrating external web-services
By enabling integration of external web-services, Agent.AI has opened the doors to make the Agents even more powerful. The Web-service interface can be slightly tricky to understand but I am sure the UI will improve over time.
Overall, my experience with Agent.AI has been memorable ❤️. I am excited to explore this platform for both my personal and professional needs.
If you liked Wander Guide, do leave a rating and review. Cheers 🍻!
Author Of article : Dhiraj Das Read full article