Building Your First Generative AI App in Just 6 Minutes — on a Slow Train WiFi

written by Stefan Christoph

November 17, 2023 - 4 minutes read

TL;DR: Six minutes on a train with bad WiFi was enough to build a working Generative AI app. PartyRock turned one sentence of description into “MarketGenie”, an app that takes a campaign objective and a target audience and returns a campaign strategy, creative suggestions, and a generated image. The interesting part: Generative AI was used to build an app that itself uses Generative AI.

Originally published on Medium on 17 November 2023, and migrated here to keep it available.

On my way back from an intense but fulfilling week with customers, partners and colleagues, I got some time — specifically 6 minutes (plus the time needed to compile the post) — in the train to make my own first experience with PartyRock, which is an easy and fun way to build prototypes and applications with Generative AI. After 6 minutes I got a running app using Generative AI to create a marketing campaign. The screenshot below shows the final outcome. But let me walk you through the process step by step.

PartyRock is powered by the wide variety of foundation models offered by Amazon Bedrock. More details can be found in the announcement blog.

I would love to briefly share the process and outcome of the very valuable 6 minutes spent.

The Goal: A GenAI App for Marketeers

Every app starts with an idea. Since I’m currently preparing my re:Invent chalk talk on using Generative AI for advertising and marketing, I decided to test how PartyRock can help create marketing campaigns from scratch. And since I’m a passionate trail runner, the marketing campaign should promote running shoes.

Building the Application

The PartyRock homepage — PartyRock homepage (screenshot by author)

PartyRock takes you into an “application builder” interface. Here, you can generate the first version of your app based on a short description you provide. Nice. Let’s try it with:

Get the Generative AI flywheel for Advertising and Marketing spinning by helping you to master the entire lifecycle of marketing campaigns with the help of Generative AI.

Generating an app from a short description — Generation of an app from just a little description (screenshot by the author)

This results in a very first version of my app, which got named “MarketGenie” and welcomes users with:

Welcome to the Generative AI Flywheel for Advertising and Marketing! This app will guide you through the key stages of creating marketing campaigns powered by AI.

Pretty solid for a first version. The interesting aspect here is that we just used Generative AI to create an app, which will itself use Generative AI to provide some awesome functionality to the user.

Trying the Application

Let’s see what the app can actually do. The first thing it asks us is to provide a campaign objective. Let’s go with:

Increase the market share for our new, super-light but rigid trail running shoe by 10% until UTMB in August 2024.

MarketGenie asking for user input — MarketGenie — the app we just generated, asking for user input (screenshot created by author)

The next question for us is “who is the target audience?” We go with:

Ambitious trail runners, running on and off wild mountain trails for long distances and duration, 100 miles and beyond.

And that’s all the input “MarketGenie” needs from us. From there, MarketGenie generates multiple artefacts. These include a campaign strategy which reads:

[..] Campaign Name: Conquer Your Inner Mountains [..] Strategy: Leverage the thrill and challenge of mountain trail running to position the shoe as an essential tool for conquering one’s inner limits. [..]

Leverage the thrill? That sounds right to me. Furthermore, it creates suggestions for potential campaign creatives. The app even proposes a first creative image which it created.

Lastly, the “MarketGenie” app even provides a chat interface which allows you to review the campaign by asking questions. Pretty neat.

Next Steps

I stop here, but you can start modifying the app by changing the configuration of the different building blocks, adding or removing elements and iterating to make “MarketGenie” even more useful. You can access this version of MarketGenie at partyrock.aws/u/stefach/sG4w0cobt/MarketGenie.

PartyRock lets you publish your application with just a click. The published version also allows others to tinker with their own version of it by generating a “remix”.

Can’t wait to see where you take this — be it for “MarketGenie” or your own use case or app. Please share what you came up with :)

Wrap-Up

PartyRock provided a super easy way to generate and compose fun applications. This allows us to test ideas and playfully explore the possibilities Generative AI can provide.

If you want to dive deeper into building Generative AI applications and are interested in how to design and build production-ready applications fuelled by Generative AI, and happen to be at re:Invent, I recommend joining me at my chalk talk on this very topic.