Three waves of GenAI “Hot Tech” power the future of work

HFS is redoubling its focus on the “Hot Tech” (formerly known as Hot Vendors) that shapes the future of work. HFS Hot Tech describes stand-out product and service providers focused on emerging tech—providers we believe will impact the future of work in the enterprise. As part of our renewed commitment to this space, we have identified three overlapping waves of start-ups that enterprise leaders must become familiar with and embrace. Each has generative artificial intelligence (GenAI) at its heart.

For each wave, we provide example firms that have excited our interest (see Exhibit 1). With one exception, Roots Automation (named an HFS Hot Vendor in Q1, 2021), none have been subject to the scrutiny reserved for and required of those we will label as HFS Hot Tech providers.

The first wave pushes beyond low code, allowing us to use natural language (NL) to create applications. The second wave enables us to use NL to interact with pre-existing apps, effectively training them to become new apps aligned to our specific needs. The third wave enables using NL to create agents to achieve the goals we assign them.

Understanding apps’ rapidly changing capabilities is critical for leading enterprises because the apps they use contain and define much of what they do.

Exhibit 1: GenAI is core to three emerging waves of start-ups shaping the future of work

Note: Company logo placement provides examples for each wave only. No meaning should be assigned to each firm’s relative positions within each wave.
Source: HFS Research, 2024

Wave 1: Use natural language to build apps

The first wave enables us to create applications using natural language prompts, sketches, and spreadsheets.

If you want evidence that this category is getting hot, look no further than validation from Google. Google Stubbs is part of its Makersuite for developers. It simplifies creating AI-generated apps. One prompt (using natural language, as most prompts do) is all it takes to create an app prototype. The relevant Stubbs update has yet to be formally released (though details have leaked online), and Stubbs, so far, is only expected to create prototypes, not complete code.

Of course, ChatGPT can deliver code snippets for an app you prompt into existence. AI21 Studio, from AI21Labs, enables natural language interaction for users to build applications.

But not all prompts have to be text. With developments in multi-modal models, drawings can act as prompts. Examples include Draw to App (which turns sketches into apps) and Tldraw, which also enables using sketches as prompts. Meanwhile, Glide allows users to create data-driven applications from spreadsheets.

Bitmagic Games provides a taste of what technology like Stubbs could mean with an offering that creates games from natural language prompts. Meta has been trialing video from prompts, but Bitmagic takes this further by generating playable games from your prompts.

Wave 2: Use natural language to interact with and train apps

The second wave enables using NL prompts to interact with pre-existing apps, effectively training them to become new apps aligned to our specific needs.

It does not take a great stretch of the imagination to see how we could use prompts to design business-centric applications. At Adept.ai, a machine learning (ML) research and product lab, early birds can request access to Adept Experiments, the first of which is workflows. Examples include opening invoices attached to emails to extract information and enter that information into accounts payable software. Another creates a custom workflow for the processes in corporate recruitment.

Adept, a multi-model AI agent, is a machine learning model intended to interact with everything on your computer. The intent is to execute actions across different software without requiring hundreds of API integrations. It can create workflows bridging various applications, effectively creating a new workflow application interacting between multiple tools. The firm already has its first enterprise customers and has raised $350 million on a valuation of $1 billion.

Rabbit combines a large language model with a large action model to understand and act on intent

Models for controlling computer actions are significantly less mature than language models, which is likely why Rabbit caused such a stir at the recent 2024 CES event with its approach to interaction across various apps. It uses hardware (its R1 device) and a large action model (LAM) to deliver this interaction. This model understands and enacts human intentions on computers. A large language model (LLM) understands what you say, and the LAM actions your request, meaning any user could teach it new skills.

We can consider a LAM as a replacement for robotic process automation (RPA). A LAM will learn interfaces from any software, delivering a new way to automate existing applications. It solves the problem of islands of apps that otherwise would not integrate with each other.

Rabbit applies the connectivity it enables to the many applications sitting on your smartphone. The R1 device has a camera and uses GPS to provide context for its decision making and actions. You can use voice to ask questions and get voice and text responses. With the support of the LAM, you can ask for “a ride home,” and Rabbit will use your preferred smartphone app to make the booking, understanding where you start the journey from.

The real breakthrough happens when you need a range of apps to solve your challenge, such as booking a vacation. The makers claim Rabbit can respond to and fulfill complex requests such as, “Book me a vacation in London for two adults and a child and find us a great hotel in a central location.” Once the itinerary is generated and reported back, it takes just one human click-to-confirm to trigger the tech to complete all the bookings required.

Wave 3: Use natural language to create agents

The third wave ushers in an era of independent bots with a mission. There are overlaps between the three connecting waves we have described, but the third wave enables us to use natural language to create agentic systems—aka agents. An agentic system can pursue complex goals with limited direct supervision (according to Open AI’s definition). It can make independent decisions, take action, and adapt to changes in conditions within constraints set by its human orchestrator.

Now, you can use natural language to set goals for digital workers

Roots Automation, an HFS Hot Vendor in Q1 of 2021, has banged the drum for digital co-workers since its launch in 2018. Now, it is preparing to launch an “autonomous workforce” platform to design, build, and direct an AI-powered digital workforce using natural language.

Designed initially for the insurance industry, the platform allows users to use natural language to describe the goal they need a digital co-worker to achieve. These digital co-workers understand and interact with the typical business applications, document types, and data found in the insurance industry.

Newo.ai also offers a platform for creating a range of action-oriented digital employees, including support assistants, sales reps, human resources associates, coaches, and tutors. Using a low-code interface, engineers build these digital employees to order. Knorket.ai brings agents to data-led decision-making and governance.

The potential for such agents is immense, offering everything from a co-pilot for the CEO to autonomous supply chain management, personalized health coaches, and AI-driven education platforms. OpenAI has launched a program awarding grants to fund research into the impact of this form of substantially independent AI and for ideas to make them safe, offering awards of up to $100,000.

The Bottom Line: Understanding changes in application capabilities is core to business success. Our redoubled focus on Hot Tech will be your guide.

Applications are how work gets done. Generative AI is driving rapid changes in how applications are built, how they can interact, and the goals they can set. You must stay close to new capabilities as they emerge. HFS is here to help. In 2024, we are redoubling our focus on emerging tech in start-ups. The three waves will play an important part as we dive deeper into what we are calling HFS Hot Tech. Look out for regular updates as we track the most relevant and identify the opportunities for impact in the enterprise.