Creativity, craftiness and clever coding defined our first-ever AI hackathon.
We’re no strangers to artificial intelligence at 24.com, but the advances in the technology have presented so many more opportunities to play. And play we did when we challenged ourselves during our first AI-focused hackathon – with spectacular results.
Teams from across 24.com came together over a 24-hour period to press pause on their normal workflow and put their talent to the test under time pressure. “They had to come up with something that could either amplify our media business – what we offer – or improve a process or workflow,” says Tobie Vermeulen, Chief Information Officer at Media24 and head of 24.com.
The mostly good, the kind of bad and the really great
The setup was hybrid – 24.com has offices in Cape Town but many employees work remotely. To make it accessible to everyone, a Microsoft Teams meeting and chat was open to the entire 24.com team and anyone could add commentary in the chat while the teams worked on their entries.
Many of the teams came into the office to connect and bond while they worked. The energy was palpable as the clock ticked – a marked difference from when the event was first announced.
There was initially little enthusiasm because of the enormous workload the 24.com team has to manage. Critical projects were underway and it seemed far too luxurious to spend a day experimenting and playing with AI tools.
“People said they didn’t have time to take part,” says Charlene Rolls, Project Head: AI. But feedback after the event was overwhelmingly positive. “One person said it had reenergised the teams to take a step back and get a breather from day-to-day tasks. I think we’ll have more people entering next year.”
The nuts and bolts
Anyone could enter, irrespective of their role – they just had to use AI tools to create their product. They could enter as individuals or teams and could request access to any paid-for tools they needed on the day with prior approval.
Anyone could enter, irrespective of their role – they just had to use AI tools to create their product. They could enter as individuals or teams and could request access to any paid-for tools they needed on the day with prior approval.
Twelve presentations were shown to the judges and it was a tough task for Media24 CEO Ishmet Davidson, Tobie, Rika Swart, GM of the News division, which publishes News24 and Netwerk24, and Charlene.
“For their presentation, they had to show what problem they had solved, what AI tools they'd used, and how much it cost. These tools aren’t cheap and we needed to see if it’s affordable when scaled up,” Charlene explains. “They also had to do a live demo, which as anyone who works in technology will know, is a risky challenge on its own. But many of the 24.com team showed up in person and online to cheer them on and the vibe was just amazing.”
The hackathon wasn’t necessarily about coming up with the next big project or groundbreaking tool, says Charlene.
“The point was to get employees to play with the available tools so they could understand what they can do. You don’t need to be a data scientist or to build models or to interact with technology in your day-to-day job to use AI.”
A purpose-driven hackathon
A core element of 24.com’s identity is to do work that matters to society. And the hackathon proved that our teams embrace this wholeheartedly.
We don’t just build products for the sake of building products. We build products for a purpose,” explains Charlene. “This is why teams were tasked with developing a tool that would improve or amplify a service offered by 24.com. All of this is geared towards serving our audiences or our clients.”
It was amazing to see how aligned our people were with this purpose, she adds. “They’re not just software engineers. They’re not just product people. They are engineers and product teams who work in the media business. And they care about the products we serve to our audiences.”
Tobie was blown away by what the teams achieved.
“The 24.com team just again proved that the calibre of talent we have is unsurpassed! I am super proud of our team!”
3 key innovations revealed at the AI hackathon
1. AI summaries – the time-saver we need
What does it take to be a winning team at an AI hackathon? For Fredrik Erasmus and Ryno Coetzee (Senior Full Stack Engineers), Eesen Changien (Associate Manager: Full Stack Development) and Nabeel Gool (Full Stack Engineer), the AI hackathon showcased their time management, courage and respect.
“It takes quite a bit of courage to try different technologies,” says Fredrik. “Everybody showed a lot of courage to come up with ideas and present them to everybody. The respect is reflected in how we recognise each other’s efforts.”
Their team initially worked on a different concept entirely before changing direction. “The most important learning was to build something that ties in with the business objectives of Media24,” Fredrik explains. In line with this, what they eventually landed on was a project to summarise News24 articles.
“The summarised content would live on a page of its own,” Fredrik elaborates. “We used Microsoft Azure Cosmos DB to store the summaries, Next.js for an admin dashboard, and our existing Nova Web solution to host the summaries page.”
Users could see a summary of each article to help them decide whether to read it
“In this way, we create additional traction to the website. We also incorporated ads into the summary page.”
Fredrik believes the value of AI summaries lies in helping us to save time. “Imagine you have a YouTube video, and you can use generative AI to summarise the content using its transcript,” he explains. “Imagine taking a report and prompting a generative AI model to summarise it for you? It will save you time
2. Customer services chatbots – there for our users 24/7
Team Artificially Intilligint – consisting of Sven Schoof (Product Team Lead), Roxanne Williams (Head of UX Design), Juan Koegelenberg (QA Team Lead) and Nico-Ben De Villiers (Software Engineering Lead: Mobile Apps) – used Microsoft Copilot to assist the Media24 contact centre.
“We work closely with our in-house contact centre,” Sven explains, “and one of the challenges they face is the sheer scale of the recurring issues users complain about.”
They devised a concept around a chatbot that would be able to resolve common user issues, freeing up agents for more complex problems.
The biggest challenge the team identified with creating this concept is getting the bots to do what you need them to do.
“It’s a reality of today’s chatbot integrations,” Sven says. “The amount of learning required to ensure that they are always contextual and understand the users is immense.”
Despite these challenges, Sven and the team believe customer service bots will become increasingly popular.
“With AI, chatbots can collect important information at the beginning of an interaction – using routing and intelligence to either resolve the issue or to relay the conversation to the best agent based on skill, availability and capacity.”
While Sven doesn't see call centres disappearing completely, chatbots can handle straightforward issues, creating a win-win situation for both contact centres and users.
3. Generative search – get the answers you need when you need them
As part of the “Waldo’s” team, Kevin Prior (Solutions Architect) and Adele Hamilton (Product Manager) – along with teammates Pieter Gerber (Head of Core Architecture), Caryn Schreuder (Head of Consumer Product), Eduard de Klerk (Solutions Architect) and Dylan Jones (Senior Engineer: Front-end Development) – wanted to investigate an area that’s familiar to many users of AI: generative search. They created a concept that would help users get answers to simple factual questions and surface relevant content.
Kevin shares: “The AI hackathon gave us a chance to become better acquainted with the actual abilities of the tools at our disposal [such as Azure AI Search services, which the team used to create their proof].”
“It allowed us to focus on sharing skills and knowledge across the different work streams, and it was a wonderful reminder of all the talent and passion in our 24.com product and engineering team,” says Adele.
They believe there is still room for improvement when it comes to generative search.
“It's only as good as the data source(s),” Kevin explains. “We think that the future of generative search lies in effectively combining multiple streams of data seamlessly to provide a richer experience.”
