Sharing some thoughts on the adoption of AI employees and what I believe to be the ‘missing interface’ required for mass adoption. I wrote this piece in 2024 and thought it’d be worth sharing.
Disclaimer: I’m not an expert on AI nor do I claim to be. These are just some random thoughts and ideas I’ve had as I’ve witnessed the pace of innovation in AI. Many of these ideas are probably commonplace lol.
My thinking is as follows:
It's becoming more clear that, in the future, a significant portion of a company's workforce will be 'leased' AI employees. More specifically, all routine tasks and work that requires very little strategic thinking will be done by AI employees.
Due to the high costs associated with developing and training these AI employees, it's probable that we see 1-3 'winners' per function. In other words, 1-3 companies will offer AI-sales reps that are better than all others; 1-3 companies will offer AI-engineers that are better than all others; and so on.
If 1 & 2 are true, businesses in the future will be run by AI employees that are provided by a handful of different companies. These employees will be expensive but less than human employees. Businesses that hire these employees will want to protect their investment and make sure they see returns.
If this thinking is accurate, it is also probable that businesses will need an interface to manage and track the performance of their AI employees in real-time. A company that focuses only on providing this interface will be able to build a better onramp to AI employees and ways to manage a multi-functional 'fleet' of AI employees than the companies focused on building the individual employees themselves.
The company that builds this interface could:
Make it easier and less repetitive to onboard new AI employees coming from different providers (i.e. all relevant company data can be automatically passed to the providers via this interface).
Enable businesses to track AI employee performance via a central dashboard vs. having to login to multiple provider's systems. This gives the business a real-time overview of their entire workforce.
Make it easier to 'train' AI by providing corrective prompts or new information to an entire workforce via a single node rather than having to go through individual provider platforms. In other words, all agents, regardless of their maker, have access to and can be trained on the same information in real-time.
Make it easier to identify issues that may be occurring in a 'department' of your AI workforce in real-time so action can be taken by a human.
I'm sure there are other benefits, but these are the main ones that come to mind off the bat.
Some other random thoughts include:
Building an interface like this could be the onramp to AI employees for enterprises and small businesses who may like the idea of AI employees but have no idea where to start. The interface is basically 'the easiest way to bring AI employees into your workforce.'
Ultimately, this interface becomes a marketplace for AI employees, making it easier for businesses to onboard, 'fire,' and 'replace' AI employees with limited downtime. If all of the relevant training data and context (from a version of AI performance reviews) is captured by & stored in this interface, new AI employees can pick up right where their predecessor left off, eliminating employee ramp up time. In other words, if there is specific context on why you fired an AI employee from company X, when you onboard an employee from company Y this context will automatically be digested by the new AI employee. This is important as the companies building AI employees are not incentivized to make it easy for their customers to switch to competitors.
You can probably also create some cool information sharing controls to box off sensitive data to specific AI employees and their providers. Right now all integrations and information sharing is managed by the employee provider, a middleman could be important to get enterprises comfortable with leveraging AI employees. This would be positioned as the most compliant and secure way to onboard AI employees or something like that.
Ultimately, this company could begin building their own AI employees, but this creates risk.
The biggest risk to all of this is if my thinking above is flawed. AI employees may never catch on and there may be only one company that wins in the long run: Open AI. Also, you'd need to build API connections with all of the employee providers. BUT...if my thinking is correct, a company like the one I laid out will be key in driving the adoption of AI employees and will likely be a must-have tool for nearly every company on earth.
I'm not sure if anyone is building something like this right now. If so, would love to connect with them at reedeswitzer@gmail.com.