SuperApps have had great success in the Asian markets, but not so much in the West. Make no mistake, the tech leaders in the US markets have the same objectives as SuperApps – connecting your friends and lenders alike, they want to be in the middle of the every transaction – emotional or financial. Yet, there are a multitude of reasons behind no SuperApp has emerged in the Western markets. As Generative AI matures, all that could change very fast.
If you haven’t noticed, the apps that dominate our daily lives have been smoothly sliding LLMs into our dm-s. Google rushed Bard out of the door when they saw ChatGPT, Microsoft has Copilots ready for every enterprise offerings, WhatsApp got Meta AI, SnapChat launched My AI, X recently got Gork, while Alexa and Siri are now the boomers in this market. Below the veil of chatbots, this has reignited the race for SuperApps. What is different this time? How would the SuperApp look? And why would this race be the last?
Chat as the ultimate UI
At the core of any human computer interaction we have, Input -> Compute -> Output. It has been a long road to reduce the friction at the boundaries of compute – making computers comprehend our intent and respond back meaningfully. For decades we have been stuck with the graphical user interface – which at best provides clunky forms with restrictive UI patterns.
But standing in 2023, that friction has been smoothened out considerably by LLMs – our natural language is very close to being sufficient to instruct the computers. On the output side, we no longer need to wrestle to make the computation results comprehensible – we can get natural language description of the compute at the right level of compression that reduces the cognitive load for interpretation.
What could the frontend of our apps look like now? Does an average white collar worker need to click though multiple tabs of internal and external websites to search and sift for a needle in a haystack? Or could they just ask for the information to be collected and summarized on their chat app, while focusing on making better decisions? Does the busy parent need to click through a between multiple tabs of calendars, email, shopping websites to figure out how to plan the week’s dinner, month’s budget, carpool schedules? Or chat on their favorite social app with a bot that understands the complex dynamics of the modern family letting the parent build fulfilling relationships?
Have you noticed how every non-conversational apps like your bank, your brokerage firm, or utility company now has a barely functional dumb chat client plugin on their app or website? There could be hundred more examples, but the pivotal example is that of ChatGPT. The most successful product launches in history – just a simple chat UI with a single textbox, and a few buttons. All the GUI elements, which are really proxies for natural language have melted away in the face of direct conversation which needs no user training.
This is why conversation first apps like Slack, Teams, Twitter, Whatsapp, Meta’s messenger (Meta AI) are well placed to become SuperApps both in the office and home front. Most consumer facing apps will evolve to become rag (Retrieval Augmented Generation) apps as they ditch the whole click driven UI to become chat first UI. Or, they will melt into one of the potential SuperApps. Just like every company was getting on the web in the early 2000s, and then on the mobile App Stores in the last decade – sometime down the line dedicated websites, apps for individual establishments could become a relic of the past. It would be much economic for many organizations to just have a Custom GPT, which can access the shared data and uses the chat interface of the SuperApp. The question is not “if” or “when” – but really, “where” will that user interaction happen? How will the SuperApp contenders need evolve their platforms to win the trust of the consumers and industries just like App Stores did a decade back?
Moving from the user interface to a lower level of computing, we have the compilers or interpreters that take pseudo english programming languages and convert them to zeros and ones for the processor. With natural language being the ui, do we really need intermediaries like a programming language, a compiler?
GPT plugins, specially Code Interpretor, have already shown us what just in time computing can look like. Stretch your imagination a little further where integration, implementation, testing and execution all gets done in real time as response to your prompt. The fact that LLMs today can generate quite accurate code at a speed acceptable to many real-time scenarios is the second big opportunity for creating SuperApps. Niche business logic that earlier took months to code, test and ship can now get autogenerated, compiled and executed – all on the fly based on the user prompt.
Dynamic code generation and just in time execution with the continuous re-authoring of complex business logic based on user feedback will be key factors in the race for SuperApps. As the landscape will go through tumultuous disruption, the pace at which we write code and ship software will no longer be sustainable. Current development best practices say all development, testing etc needs to be moved left. But my guess is that as LLMs get better, we will see an absolute paradigm shift where the entire software development life cycle moves all the way to the right, towards the edge. The models in the SuperApp will eat and regurgitate the software stack into a completely new beast that focuses on quality, safety and security.
Data – the only moat
With the user interface and computation leaving less room for competition, it all comes back to the age old wisdom: the one who keeps control of the data wins. The absolute key battlegrounds that every app is fighting for and building moats around is data. Both historic data and real time data. In the last decade we saw how tech companies begged, borrowed and stole user’s data to build billions with personalized recommendations. This gives them massive advantage in terms of training their closed source models with all forms of human interactions. That base paves the way for Copilots or CustomGPTs to be the next generation of personalized recommendation.
As these tools supercharges content creation on these platforms, so will the monetization of it. GenZ knows this. They are way ahead of the curve in generating content on these SuperApps contenders, that celebrate and reward their unique perspective to captivate attention. They know media entrepreneurship will be the most common profession of their generation. They know their data will work for them while they sleep. The question here again is, which platform will they choose with their data and content?
The last race
On the spectrum of market capitalization, network effects, advanced models, partnerships, and product offerings, contenders like OpenAI, Twitter/X, Google, Meta, and Microsoft vie for supremacy. Each brings unique strengths to the table, making the race for SuperApps an exhilarating competition to watch unfold. However, the competitive advantage that each of the players has, could be disrupted in a matter of days obliterating empires. Whoever has the correct combinations of technology and market fit, will attain escape velocity before others can catch up. Case in point, the never before seen user acquisition by ChatGPT – 1 million users in 5 days. Recall this February, Google’s drop in share prices by 9% after Bard’s demo had mistakes in the response, home work helper Chegg losing 50% of value due to ChatGPT’s four months of existence. We have no clue on how to regulate the power that hides behind the innocuous interface as it flywheels into a monopoly.
It has become clear that the success of the AI arms race is biased towards deep pockets. Our only hope is that in the past, we have been successful in building open technologies that benefit the whole civilization rather than for profit companies. Take the Internet for example which triumphed over the proprietary Information Superhighway that Microsoft wanted to establish in the 90s. Open Source softwares like LAMP stack that got the world to its digital adolescence. Or Bitcoin or Ethereum are hallmarks of decentralized technologies. We need open standards, protocols and software for building and regulating models. That way when singularity unfolds, it does not happen behind the closed doors of multibillion dollar corporation, but out in the open with the whole humanity at its core.
With proper regulation, open protocols, open weights, may be we have a chance that we don’t end up with the dystopian future that with AGI. We will not be in awe of harnessing it. We will not be bothered about “where” it comes from, “how” to make it work with the world around us but take it for granted that it works for us. Instead, computing will fade into the background allowing us to maximize our unique gifts to build something new.