Orchestration: The Next Front Line for AI
While much of the current focus of generative artificial intelligence (GenAI) and other AI techniques is on delivering functional use cases, the next horizon will bring something more exciting and all-encompassing: AI orchestration of processes and experiences for both the individual and the enterprise. Kelly Goetsch, Chief Strategy Officer at commercetools, explores the future state of AI and the role of retailers and composable vendors in this new paradigm.
Imagine a world in the not-too-distant future where you’re actually having a conversation with an AI-powered app. You can ask it to show you the closest electronic retailer that sells a specific phone, have it find you Apple Air Pods online for 30% off or make it locate a highly rated eBike under £3,000 that would fit in the back of your vehicle. While AI-powered shopping solutions that can deliver this experience aren’t too far off, there are a few issues that need to be addressed before they become a reality. First, many retailers aren’t able to feed AI tools with up-to-date data because they don’t have the right technology in place. Second, when retailers try to publish new products, it can take months for common language learning models (LLMs), like ChatGPT, to update their datasets and then actually return results to potential customers.
To make AI-powered apps a viable eCommerce channel, customers and retailers need data updates to happen in real time. Once retailers can expose data instantly to AI tools, they’ll be able to surface products, availability, pricing and promotions very quickly and easily and increase business value. By taking an API-first approach and leveraging composable architecture that enables different systems to deliver each business capability, retailers can take advantage of AI at the same time consumers turn to it to meet their needs.
The question will then become: Where will consumers turn to ask their shopping questions? And how can your company be ready in this brave new world?
The Battle of Personal Assistants
While personal assistants from some major players may have a lot of your data, ChatGPT doesn’t have easy access to personal information just yet. This means there’s still a long way to go for personal assistants to unlock the value of AI while facing several deficiencies along the way.
To overcome these challenges, the ultimate personal assistant in the next generation will be powered by open AI APIs.
“AI is an absolute 100% necessity for the ultimate personal assistant.” Kelly Goetsch, Chief Strategy Officer, commercetools
But can the big players catch up and truly be able to leverage our personal graph? Will consumers give them permission in accordance with the Guidance on AI and Data Protection and other regulations? Will they be able to browse through our email purchase history to understand which products we have and have not bought and then apply all they know about us to provide the best experience? Will they have access to the systems needed to take actions on consumers' behalf?
The underlying language models have been figured out and are all usable by now, with AI largely commoditized and cheaper than ever. Just as the cost of DNA sequencing has dropped from millions of dollars to $10, the same is happening with AI. This level of accessibility means that players in the space will all be able to easily pick and implement the AI tool of their choice to empower their assistants.
Since so many language models are readily available for use, the true differentiation between personal assistants will come from the data within the ecosystem: the data the assistants have on you — first-party data – and how effectively it can be leveraged.
Therefore, the winning personal assistant will be the one that can:
- Understand your personal graph
- Have the greatest legitimacy and connectivity across composable ecosystems to undertake actions and provide assisted services to the consumer
- Gain consumer trust to utilize personal data across the shopping journey
A New Paradigm for Retailers
Streamlining the customer journey through AI assistants has many upsides for consumers, but how far are retailers going to allow such assistants to transact on consumers behalf using your API? Are they going to be happy moving to the back of a transaction process? Or will they throw away their API walls to make sure they get the consumer and their all-important data?
Many blue-chip brands started selling products on Amazon because it disrupted the market. Amazon commoditized the experience, and it also made it a lot faster to buy, browse and ship. Brands realized that if everyone is on Amazon, they must be on Amazon, too.
AI and retail will go much the same way. In the future, there will be a split responsibility for the transaction process: UI-driven experiences and API-to-API interactions. For high-value purchases and luxury retailers, brands will want customers to spend time in a brand experience site to help justify the prices. For commodities, however, there will be a growing emphasis on API-to-API interactions, with programmatic buying and selling of products. Like Amazon, a similar type of marketplace will likely emerge for AI assistants to interact with. The benefit, in this case, is that when retailers have all their commerce APIs open to the world, they’ll still own the customers and be able to remarket to them.
The balance between UI-driven experiences and API-to-API interactions will vary greatly by category, depending on the consumer awareness and understanding of the products. We’ll see very different brand-building and marketing mixes to power consideration and support a new AI-powered paradigm.
From the Personal to the Enterprise
The impact of AI assistants will go beyond the personal user. Businesses are going to look to GenAI and AI to run multiple processes and systems of record. They’ll use AI to help ship tens of thousands of a particular product and enable everything from demand generation to pricing and promotion; the whole journey from discovery to loyalty.
Some processes are more effective or efficient than others, and clearly it's not beyond the wit of man to think about how you could orchestrate across those processes to drive business outcomes leveraging generative AI.
The role of enterprise orchestration is the key front line for AI, as we move into a future with increasingly sophisticated and heavily orchestrated multiprocesses. In this future where orchestration can be powered by the richest possible context and highly domain-specific, businesses will look across everything from retail, buying processes and fulfillment to processing and warehousing processes to drive optimization.
Who will drive this future? Who has the greatest position in the marketplace to deliver on AI enterprise orchestration? Who is best to own the data? If you’re a retailer, who do you believe most understands the retail domain and is most capable of orchestrating multiple retail processes front to back, organizing the data and providing the greatest assisted outcomes from a core perspective?
Gone are the days when digital platform vendors tried to bake all known capabilities into a single, monolithic product, incorporating integration platforms as a service (IPaaS), analytics, personalization, search and everything else. The whole pitch was that they’d do everything for you.
In a composable landscape, there are different roles for different vendors to play and different use cases for AI. There’s content authoring in content management systems (CMS) product information management (PIM) platforms and digital asset management (DAM) tools and GenAI accelerating content generation. For fraud and payment capabilities, there are now specialized AI solutions that can make smart decisions.
With pure commerce capabilities, however, there’s less of a case for using AI directly. For example, commercetools has a relatively narrow surface area, but it goes deep and exposes data and functionality through clean, well-documented, publicly available APIs. As a result, commercetools’ domain is in building solid APIs that enable companies to retrieve data in real time, but it doesn’t play an active role in AI orchestration.
What businesses can expect of a composable commerce platform like commercetools in the AI landscape is enabling easy, fast integration of AI-based solutions across the commerce journey from AI-powered search to inventory management.
"You shouldn't have a specialized AI team, where you're explicitly going out to build AI features. Instead, you should build features, period. And if AI happens to benefit the development of the feature, it should be used behind the scenes." Kelly Goetsch, Chief Strategy Officer, commercetools
The role of orchestration will likely fall to software vendors who already act as the glue between multiple other systems and to new startups whose only job is to deliver AI orchestration.
Preparing to Take Advantage
The pace and direction of change is accelerating. While we can only hypothesize about what the mid- to long-term looks like, what we do know for certain is that the ability to respond to change quickly is imperative. Modern, modular, scalable, composable architecture, that provides access to clean, well-structured data and enables best-of-breed capabilities, is the perfect precursor to GenAI and AI orchestration. You can start your path to transformation by adopting a microservices-based, API-driven, cloud-native and headless (MACH) architecture as the technical backbone of your composable commerce. Combining an array of modern technologies, MACH enables more flexible, scalable and agile commerce, meaning that the time to ditch the monolith is now.
"Composable is the ‘what.’ MACH is the ‘how.’" Kelly Goetsch, Chief Strategy Officer, commercetools
Taking Advantage
EPAM and commercetools are uniquely positioned to help you take advantage of this wave of disruption and maximize the benefits of AI-powered commerce adoption. We can help you turn your vision for GenAI-driven commerce experiences into a reality, leveraging our engineering excellence, end-to-end AI capabilities and commerce expertise.
