From search to prompt: Consumer AI will eat the app world
While we’re still early in the AI adoption curve, there are already signs of scale
Artificial intelligence (AI) isn’t solving world hunger (yet). However, one can’t deny that it’s all around us.
The GenX-ers in my family ask ChatGPT anything from what dish to cook to which stock to buy. A millennial friend has built an AI agent to automatically apply to all jobs that fit his criteria and personalise outreach messages. Gen Zs around me use it for selecting what outfit to buy, helping with their college assignments, and even as their personal therapists.
There are two observations from all this: one, AI adoption is real and there’s incredible latent demand, and two, ChatGPT is becoming the new Google. A study by OpenAI reports that almost 80 per cent of all interactions with ChatGPT is for practical guidance, seeking information, or for writing help. Web 2.0 was built on search, and the AI world seems to follow the same trend.
Our part of the globe has, of course, adopted consumer technology en masse over the years. South-east Asia and India contribute significantly to user bases of the largest consumer apps, and the tailwinds have never been stronger: massive populations that have been brought online in the prior technology waves, affordable data and smartphones, and 25 to 30 per cent of the world’s Gen Z.
The question has always been monetisation. However, that has changed considerably in recent years.
India has seen some scaled up consumer applications across content (KukuFM), astrology (Astrotalk), personal assistants and productivity (Truecaller) and more. South-east Asia’s decacorns, such as Grab and Sea, power several kinds of consumer transactions. All these are built on revenue. Many global consumer leaders are building their overall growth plans based on tailwinds from India and South-east Asia.
BT in your inbox
Start and end each day with the latest news stories and analyses delivered straight to your inbox.
While we’re still early in the AI adoption curve, there are already signs of scale. Both India and Indonesia are among the top contributing countries for ChatGPT usage.
Almost all AI leaders have reduced pricing for India to drive further adoption. The cheapest ChatGPT tier, Go, is priced at roughly US$4.50/month, compared to the cheapest US plan, which is US$20 a month. Perplexity and Gemini have offered their pro subscription for free through partnerships. Seamless and free-to-use subscription payment enablers, particularly Autopay in India’s Unified Payments Interface stack, have made these more accessible.
In the Internet era, search unlocked a wave of vertical platforms. AI is now triggering the same shift. Over the last 18 months, there has been a Cambrian explosion of AI-enabled consumer platforms across most sectors one can think of, completely reimagining software and platforms for an otherwise existing demand base.
Many of us shop online on Shopee and Amazon, watch content on Netflix, listen to music on Spotify, learn courses on Coursera, book trips on Traveloka and MakeMyTrip, use StackOverflow to write better code, and so on.
Each of these is now seeing AI-fication in different ways, some of which could break out and change how we transact for good.
Globally, education has been a large contributor to AI search volumes, so much that ChatGPT sees substantial drops in usage around US school/university vacation months. India has the highest Gen Z and Gen Alpha population in the world, with education and ambition ingrained societally in them.
There are comparable numbers of Gen Z and Gen Alpha individuals across all Asean countries who also have not been able to access the best education to progress in their academics and careers. Even post the first EdTech wave, we see similar issues continue with education: tutors are expensive and don’t always have time, learning pedagogy is standard and not customised, and current educational methods are usually one-way and not interactive.
AI applications solve most of this with being accessible at the learner’s convenience, can customise curriculum and content to the learner’s needs and pace, and are as interactive as it gets.
At work too, there is a slew of productivity apps, often termed “copilots” and “agents”. Some 25 to 30 per cent of code is now written by AI in leading companies, and this is a fast-growing phenomenon.
A growing share of content on YouTube, Instagram, and TikTok is AI-generated, helping influencers grow their following faster. Similar tools are taking off in slide generation, music creation, process automation, and more, initially being adopted by employees in companies in a personal capacity to help them work better, and subsequently through enterprise-wide deals.
One of the largest potential markets out there is companionship (in multiple forms), and we see AI companions on the rise. US and China-based AI companion apps that enable users to speak to their favourite celebrities or characters, a friend, a spiritual mentor, an AI therapist (and even an AI boyfriend/girlfriend!) have hit the top charts on app stores.
We’re excited to see how localisation could enable companions to scale in multiple formats. There are multiple things to watch out for here, of course, ranging from the long-term societal impact of interacting with AI without “human” connections to potentially unclean conversations and liability risk if the AI companion is not moderated.
Finally, we’re seeing early tailwinds of consumer AI tools to disrupt commerce. AI shopping assistants help you shop better by searching for an image of an outfit you like, finding more affordable options, enabling virtual try-ons, and more.
AI travel agents help you plan your trip from reels you like, automatically book you the cheapest, most convenient flight, and possibly even help you identify which credit card or loyalty programme to use to maximise reward points. AI nutritionists and dieticians could help you identify your regimen and help buy the right products and services that align with your goals.
Of course, the large AI research labs are churning out new models every day that could be consumerised, the latest one being Google’s Nano Banana for image generation. With further research and costs continuing to go down drastically, consumer AI apps could leverage these models and gain virality.
However, there is a risk of the incumbents innovating and launching consumer AI features within their platform. For example, Amazon just launched Rufus to enable AI-assisted shopping, and Expedia launched a trip planner from reels. Startups in this domain should closely monitor incumbent activity.
What else could help consumer AI apps win? The opportunity is a land grab and it’s all about distribution, as it was with consumer apps and social media.
Products need to have built-in virality loops (potentially social), be hyper-personalised with consumer niches, and offer a better experience with AI than the incumbents.
Cracking costs and unit economics could become a moat too; those who can optimise and deliver the cheapest AI models with an extremely intuitive user experience could potentially make the LTV/CAC equation make sense.
We are seeing this with text and image-based models, but it is yet to be seen at scale with voice, video, and other formats. Models built on these are already enterprise-grade but are only affordable for B2B use cases at scale, not for B2C (especially in emerging markets) yet.
As we continue to see consumer AI startups flourish, we’re excited about how AI-native products capture the next AI-native generations across these domains and more.
The writer is associate director, investment at Vertex Ventures SEA and India
Copyright SPH Media. All rights reserved.
