By Charles Fei and Jabran Muhammad | Senior Specialist Leads Deloitte
23 May 2024
GenAI 1.0 is coming of age.
GenAI 1.0 has seen remarkable adoption across the board, and equally impressive, if not more, has been the pace of evolution. In just 24 months, the field witnessed a whirlwind of activity, with hyper scalers racing to offer Large Language Models (LLMs) and related technologies as a service at subsidised rates. This wave of enthusiasm was not limited to the tech giants, as enterprises eagerly embraced the potential, commissioning PoCs to explore and validate the possibilities. The results were transformative, even converting hardened sceptics into advocates.
Today, most enterprises are preparing to operationalise their GenAI projects within the next 12 months.
"Deloitte study: eight out of ten global leaders expect GenAI to transform their organisation within three years."
GenAI 2.0 - Actionable Generative AI beckons
You may already be using ChatGPT or Gemini to help with itinerary planning. They are fantastic at churning out detailed itineraries and options, but you are still doing a lot of heavy lifting, like making the actual bookings, sorting out travel insurance, and planning out tailored experiences.
Fast-forward 24 months, and your trusted AI assistant might help you book your next holiday, not just churn out an itinerary. If you are looking for a two-week holiday in Japan for under $5k, your virtual agent will suggest accommodations and flights and handle bookings, travel insurance, and a visit to Arashiyama’s Bamboo Forest.
If you thought LLMs were a game changer, you'd be blown away by LAMs. LAMs, also known as AI agents. AI agents open up an entirely new set of possibilities, where GenAI is no longer a passive tool to generate text or images but more of a partner or assistant who can get things done in real-time.
Hard to believe? Read what the godfather of AI has to say:
“I think AI agentic workflows will drive massive AI progress this year — perhaps even more than the next generation of foundation models.”
- Andrew Ng
AI agents love APIs
APIs have been proven as the de facto language for machine-to-machine communication. AI agents will rely on APIs as the front door to an organisation's business capabilities. Just as websites are the digital front door of organisations for humans, API catalogues will be digital front doors for AI agents. Many lessons learned from the internet's early history will become relevant again. For example, just as SEO-friendly websites became a differentiating capability, AI agent-friendly API portfolios will materially improve top-line growth.
“APIs have emerged as the 'digital front door' for organisations looking to leverage AI quickly and effectively. By enabling fast and seamless data exchange across varied systems, APIs amplify AI's capacity to learn from multiple data sources. In a nut shell, they act as the vital link between AI as a new technology, with practical business applications, ultimately driving innovation and securing a competitive edge for organisations."
– Andrew Dent, Senior Director, Field CTO, MuleSoft
First mover advantage
The similarities between the current AI and the previous SEO era go beyond the surface level. In the same way that web crawlers used to index and favour websites with a user-friendly interface, AI agents will also prioritise businesses with a well-organised API portfolio. It will be expensive to outrank the market leaders, never mind the opportunity cost. However, being the first mover in this space can have significant upsides with little to no downsides.
AI agent-friendly APIs
AI agent friendliness is a developing field. Some aspects of agent friendliness can be added, but others must be integrated into the API architecture. For example, some quick wins teams can gain to make their API portfolios agent-friendly include:
AI agents will also rank APIs based on response time, uptime, security, and streaming capabilities; these quality attributes are an outcome of your API architecture and, therefore, are more challenging to bolt on after the fact.
“Our clients with the most mature API capabilities are talking to us about AI agent friendliness assessments, and we fully expect this to become mainstream over the next 12 months”
- Damian Harvey, Partner Deloitte.
Closing thoughts
According to Gartner, $297 billion will be invested in Gen AI by 2027. Enterprises that expose their business capabilities as agent-friendly APIs will benefit better from the value created. We’re not far from having AI agents suggest and then act on our holiday bookings, insurance quotes, and many other aspects of our lives. However, for your organisation to participate in this GenAI economy, you must invest in refreshing your API portfolio. This time, the technology is mature, talent with battle scars is available, and billions of dollars are getting invested in creating a new category of API consumers, AI agents. The time to refresh your API strategy is now.
Subscribe to the Techweek newsletter for updates straight to your inbox:
6 Nov 2024
As we announce 'Lifting Productivity' as a key pillar for Techweek25, we're focusing on a critical question: how can technology help New Zealand businesses work smarter, not just harder? With…
6 Nov 2024
When we talk about 'Creating Good Tech' – a key pillar of Techweek25 – we're not just thinking about clever innovations.
6 Nov 2024
As we announce 'Developing Careers & Skills' as a key pillar for Techweek25, the timing couldn't be more crucial.