A company's output depends on people power, limiting what can be accomplished by the size of its workforce. Breakthroughs such as computing and access to the web have driven significant productivity gains, but these advancements are limited by the same common factor: headcount.
Tools require users: The spreadsheet begets its analyst. The slides app begets its designer. Technology may become easier to use, or more powerful, but it has always required some core relationship with a human to function.
When Open AI released ChatGPT, the transformative power of AI to augment human capabilities was immediately obvious. This sparked a huge wave of excitement about 'copilot experiences' where the AI, just like a spreadsheet or slides app before it, could make someone even more efficient at doing the same work. While we share in this excitement, at Relevance AI, we view copilots as an extension of an outmoded way of working.
Copilots by their nature serve as tools that enhance an individual's productivity. We have become accustomed to this from technology over the last 2 centuries. In the same vein, most people are seeing AI as a sort of copilot, with a user as the “pilot” still driving the ship.
But generative AI is different: it is going to shift the world to an entirely new paradigm. The next great leap in progress is autopilot - the ability of AI to operate autonomously and iteratively without continuous human interaction. This new technology for the first time frees teams from the constraints on productivity imposed by headcount, making access to compute the only limitation on their output.
We define the output of teams in the following way:
- Team outputs 1 unit of work, with the limiting factor being headcount.
- Team + Co-pilot outputs 2 units of work, with the limiting factor being headcount + efficiency gain of co-pilot.
- Team + Auto-pilot can output ∞ units of works, with the limiting factor being compute.
What does Autopilot look like in your organization?
Autopilot is a fleet of AI co-workers equipped with tools and trained with specialized knowledge across different functions. These agents collaborate to solve complex tasks they’ve been assigned, and make up an AI workforce that functions as an extension of your team. The reason compute becomes the limiting factor is that the AI can be scaled up at the press of a button. This AI agent is smart enough to handle, on its own and without live supervision, unexpected responses by coming up with novel solution pathways.
It is important to keep in mind that these are not fully autonomous or conscious entities and therefore a very substantial existing human workforce is still required. It is the team's leadership that defines the objectives and results that matter to an organization. It’s the subject matter experts that distill their current and future expertise into the agents to enable them to function more and more autonomously. It is the human managers that steer the AI agents when they become confused or make a mistake; that create and apply approval processes to uphold privacy and continuity and safety and the like.
For instance, an engineer typically undergoes a code review by another engineer before publishing. A new sales rep's personalized emails to prospects are approved before messages are sent. The difference is that with an AI workforce, agents can be given discretion to complete tasks autonomously or discover a novel solution rather than just assist the human in their flow. And they learn as they go, improving their performance and execution as they are rewarded for their successes and penalised for their mistakes.
How does an AI Workforce change the nature of work?
Let's imagine a customer support team starting with one person handling 100 tickets each day. As the workload grows, the company hires more people. When the workload increases tenfold, they might upgrade their software or equip reps with a co-pilot, scaling each employee’s capacity to address 200 tickets per day.
While individual employees are more productive, the ability to complete a growing number of customer support tasks remains tied to headcount.
An AI Workforce changes that equation. Support teams need only scale compute to process ever greater volumes of tickets. Reps could manage agents that are processing 10k tickets per day.
This model will unlock the next step function change in productivity. As this sort of nearly unlimited scalability begins to ripple through the global economy trillions of dollars of value will be created by companies equipping their teams with autopilot.
Relevance AI is the platform to train and deploy your AI Workforce. If you’re a growing, future-thinking business that wants to benefit from this change in model, let us know. We’re onboarding customers with Sales and Support agents today to help escape previous constraints
