30 May 2024 12:20 am Views - 167
The PickMe Hackathon Team
The Hackathon Winners
What is Gen AI? It’s a category of artificial intelligence systems with the focus on learning the patterns of existing data at lightning speed, and using that knowledge to generate new things that follow those same patterns. It has many potential applications, including creating art, designing products, and even scientific discovery. Before the age of digital data, everything was manual and guess work, which made processes laborious and cumbersome. With digitalisation things began to slowly change and processes became smarter and more efficient. But now we have arrived in the Gen AI age, where the data or content produced over the years and stored in digital format has become an extremely useful pool of information. With Gen AI, we can use that data to make life changing decisions in our day to day activities in real time.
Roshan (not his real name) is a delivery guy for a restaurant in Colombo. He is from the Kurunegala district, and has come to work in the big city, he is still not very familiar with the streets. But in his job, he has learned some basics to complete his tasks efficiently. The biggest plus is he is courteous, an important trait in his line of business and more so because many a time, he has to ask for instructions from customers. He is polite when he calls for directions, well dressed and courteous when he arrives and interacts with customers with a pleasant smile. He is clearly well aware that positive customer ratings are key to the success of his job.
Evolution of Service -Traditional vs. Digital
The kind of service these riders offer is actually of the old world system before digitalisation happened. While it still continues in a small way and has a definite appeal that may charm the customer, this method of servicing is still primitive, in terms of fueling the city’s economy. In today’s digital world, everything must be scalable in order to find efficiencies that connect to lowering operational costs and maximising profits, a formula that would be a win win for both customers and the service providers. That’s why ride hailing, and app based food delivery services are becoming more sought after by both customers and suppliers. In Sri Lanka, PickMe has been operating their app based hailing and delivery platform for the last 10 years, and their home grown formulas have come of age. From simply formulating algorithms, the company moved into AI solutions and is now transitioning to the era of generative AI (Gen AI).
Understanding Gen AI
What is Gen AI? It’s a category of artificial intelligence systems with the focus on learning the patterns of existing data at lightning speed, and using that knowledge to generate new things that follow those same patterns. It has many potential applications, including creating art, designing products, and even scientific discovery. Before the age of digital data, everything was manual and guess work, which made processes laborious and cumbersome. With digitalisation things began to slowly change and processes became smarter and more efficient. But now we have arrived in the Gen AI age, where the data or content produced over the years and stored in digital format has become an extremely useful pool of information. With Gen AI, we can use that data to make life changing decisions in our day to day activities in real time.
Today, data is vital for businesses and organisations, it is considered the ‘new gold.’ Companies can leverage data to understand customers, improve products, and make informed decisions. With Gen AI, individuals and organisations can go further by analysing vast datasets to identify trends and patterns invisible to the human eye, allowing them to make data-driven decisions with greater confidence
Challenges and Solutions at PickMe
A Senior Business Intelligence Analyst at PickMe, Sheelu Perananthan says Gen AI can vastly circumvent many challenges the company faces in managing huge volumes of data. “Every business has the challenge of data cleanliness. We work with more than 6 million records per day, which is a really large amount of data, which if handled through legacy methods such as analytical tools and manual interventions could end up consuming many hours of productive time.
Thus, Gen AI will help us to make such processes really efficient and effective. As we operate in a realtime multi-party marketplace, we need real-time solutions to real-time problems. With Gen AI’s power and support, I would be able to use tools like Gemini, which can analyse and solve problems quickly, almost in real time. Therefore GenAI enabled solutions accelerates our speed to market, both from a business intelligence as well as from an operations point of view.”
According to Roy Ian, a Data Architect on PickMe’s Data Science Team, the company had until recently, used legacy methods to handle their time consuming manual inspection interventions such as fraud detection, driver registrations etc. “Now with more drivers joining the network as we scale the operations, we are gearing up our approach to handling such processes through intelligent and automated systems which will give us the right level of agility and accuracy. As opposed to how we used to handle registration related documentation processes such as validations, approvals, and data retrieval, now we have deployed new efficient systems using Gen AI which will actually scan all the documents such as licence, insurance etc. This allows us to check authenticity of the documents at scale, and prevent forgeries. This even extends to checks such as face recognition through selfies as well as driver’s licence match and verification of the vehicle. Our next step is to extend doing this with the Department of Motor Traffic, but that’s in the future. We also plan on checking the NIC’s with the Department for Registration of Persons”
Next Gen in data crunching
Isira Perera, COO of PickMe also says that Gen AI has the potential to streamline their processes. “In the future, we could use Gen AI to improve how we match drivers and riders with ride hailers, as well as those using our food delivery services. Right now, we rely on customer feedback about drivers through ratings and comments. This helps us categorise drivers based on personality or preferences (talkative drivers, drivers who prefer short trips etc.). The goal is to eventually use this data to match commuters with compatible drivers automatically through Gen AI analysis. Call centre optimisation is another area where Gen AI could be beneficial. By analysing customer behaviour patterns, it would identify customers who might require special handling due to their personality. For instance, if a customer frequently complains about chatty drivers, Gen AI could route their call to an agent who can better handle a more talkative driver. While this isn’t available yet, training Gen AI based on past interactions could make it a reality.
Developing Gen AI responsibly is crucial. We need to avoid bias and ensure fairness in our Gen AI-powered processes. Additionally, training generative AI for customer behaviour analysis might necessitate centralised databases with user profiles, which raises privacy concerns that we need to address.
Overall, it has the potential to transform our ride-hailing service by enabling automatic driver verification, personalised driver-rider matching, and optimised call centre experiences. However, we must carefully consider the ethical implications and privacy concerns that come with such advancements,” adds the COO.
“Looking at the larger canvas, when we are deploying various tools to give us solutions, we are finally looking at how accurate we can get in predicting customer behaviour i.e. the driver or rider who is our direct customer and secondarily, the user of our platform to hail taxis or other services. Gen AI is the next generation in the process of managing huge numbers of data which is humanly impossible to collate in real time, which is a necessity for making predictions on customer behaviour. The best example of this would be matching commuters and drivers in an environment where weather patterns are changing. As the data keeps coming in and PickMe’s operations have to make course corrections to ensure a proper balance between supply and demand, we cannot have the analysed data coming to us the next day. We need it right at that moment. This is where Gen AI comes into play, as the speed of putting thousands of bits of information into its system and generating an analysis can be done instantly. That’s why it is a very powerful tool and it will help us to add value in giving competitive prices to the ride hailer. It will also enable us to provide opportunities to our network of drivers, riders and merchants,” says Jiffry Zulfer, Founder CEO of PickMe.
Integrating Gen AI
To ensure more successful engagement with Gemini, the Gen AI tool of Google, PickMe held a 2 day hackathon recently, which was participated by their staffers from various departments such as account management, engineering, Business Intelligence, Data Science, Quality Assurance, and Product. The hackathon was led by Searce India, a partner for Google.
To maximise the understanding of Gen AI amongst participants, PickMe integrated technical presentations from Google Cloud into the event schedule. The Gen AI infrastructure and services were also provided by Google. Regional representatives from Searce who were present as Gen AI resource personnel assisted developing teams to ideate and build their applications. Solutions were provided by 5 teams with 3 winning projects selected by an expert judging panel. The team who came first, came up with an intervention related to driving better discoverability and experience in the Sri Lankan food marketplace.
Ankit Sinha, Vice President of Searce India who attended the hackathon said that the winning team stood out because they were able to articulate their business case quite precisely, so that everyone was together on the problem definition. “Another thing this team did extremely well was, they used the technology sensibly and showed proof of how the technology was solving things, which classical i.e. non Gen AI tools were not capable of. They also came up with a roadmap. They will essentially go into production in the real world.”
While PickMe plans to put the winning Gen AI solution into use soon, the judges also agreed that the other 4 solutions developed by the respective teams also had merit with the potential to be tweaked for real world use.
Comprehensiveness of the proposed solution in terms of viable proofs of concept, if it was realistic, and how implementable it was within a timeframe, the extent to which the solution upgrades customer experience, financial resilience such as saving money, and how well Generative AI tools were used were all part of the judging criteria.
The teams were also required to outline if their solution could have been achieved without AI technology.
Benefits of Gen AI and being ‘grounded’
Expressing views on how much businesses can depend on Gen AI, Ankit Sinha says it helps companies to stay ‘grounded’ in their enterprise truth.
While Gen AI must be fundamentally rooted in truth, the current focus across the globe is to ascertain the veracity of Gen AI models, which necessitates providing links to credible resources. When it comes to enterprise applications, it’s crucial to ensure they are aligned with the truth specific to the company’s data sets, rather than solely relying on internet sources. There are two levels of truth: references from the internet and grounding in enterprise truth, which relies on company data. This distinction is significant, given the divergence between the two. The emphasis on grounding is particularly vital as Gen AI’s creative capabilities are increasingly used in tasks such as summarising and reasoning.
Google develops tools that enable companies to ground their solutions, offering a foundation based on internet truth, which companies can refine according to their own truth. Ankit Sinha is of the view that, while every CEO must look forward to the potential of Gen AI, it is also essential to ensure the responsible deployment of it.
“I am very confident PickMe will be able to take over this technology. Some of the things that’s working for PickMe is that it’s a strong brand that has user attention on the application. With the kind of technology that now exists with Gen AI, PickMe will be able to accelerate its ability to understand their own users better. To be able to customise the apps better for what the users are potentially looking for. So companies with strong foundations and strong brands and strong user-loyalty are obviously going to benefit a lot more because all they need to do is keep personalising the experience for every user. This is going to be a multi-year investment from PickMe’s perspective and they will keep seeing quick returns of this technology. And that’s the other beautiful thing that’s happening with tech overall, because the cycles are not six months, 1 year, or 3 years. People generate things within 24 hours. They will not stop here and they will see returns very quickly,” says the Vice President of Searce India.