
This is Part 2 of our AI Teammates series. Part 1, which covers foundational concepts and design choices for AI teammates, can be found here. In this series, we explore various topics in an AI teammate's life cycle through the lens of Emma, an AI customer support teammate at a fictional airline.
How to “Hire” an AI Teammate
Hiring a human employee like Emma would be straightforward: write a job description, post it, screen and interview candidates, and then onboard the chosen one. Hiring an AI teammate, however, is more complex.
Where do AI teammates come from?
In today’s market, there are two primary paths:
- Buy Off the Shelf: Companies can purchase pre-trained AI teammates from vendors (e.g., Ema). These AI teammates come equipped with predefined workflows and integrations for specific roles. This process resembles vendor selection more than traditional recruiting, with the vendor largely responsible for the AI teammate and the company making minor customizations.
- Build In-House: This path offers greater control over the AI teammate’s identity, behavior, and alignment with company culture. However, it demands more internal AI expertise.
Ownership and Collaboration: When building an AI teammate, ownership is crucial. Who defines the requirements, determines its integration within the organization, and sets its KPIs? An AI teammate’s “hiring” will likely be a cross-functional effort. Front-line managers identify the need, leadership prioritizes and resources it, and IT evaluates vendors or supports internal builds. Just like human hires, AI workers are most effective when their hiring is intentional, collaborative, and based on clear expectations.
How Many AI Teammates Are Needed?
The number of AI teammates a company needs, or “how many Emmas,” depends on the specific function:
- Repeatable Workflows (e.g., Customer Support): In roles with mature processes, clear handoffs, and predictable edge cases where speed and scale are paramount, AI teammates may soon outnumber their human counterparts. For instance, five Emmas might handle daily rebookings and refunds, with one human, Luis, monitoring quality and handling exceptions.
- Creative and Judgment-Based Roles (e.g., Marketing, Strategy): These roles will likely retain a human core. On a five-person marketing team, two or three AI teammates might support content generation or A/B testing. These AI teammates function like fast, tireless, and helpful junior staffers, operating under human guidance.
As AI teammates become more capable, these ratios are expected to evolve.
Deployment: Temporary vs. Permanent AI Teammates
Beyond headcount, the next question is how AI teammates are deployed: temporarily for specific projects or permanently embedded within functions or across the organization. Both models are emerging:
- Temporary Model: Similar to digital contractors, some AI teammates are brought on for short-term projects or seasonal spikes and then offboarded. Examples include an AI FP&A analyst activated quarterly for earnings reports or an AI management consultant spun up for a discrete strategy project, trained on context, and then archived. These teammates offer scale without continuity, are hired on demand, and decommissioned when work is done.
- Permanently Embedded Model: These AI teammates become a continuous part of the daily work rhythm. Emma, as an AI customer support agent, exemplifies this, working full-time on the frontlines, handling calls, adapting to edge cases, and internalizing policy changes. Like human employees, she improves over time.
Cross-Functional Model (AI Managers): A third model for AI integration is emerging: the cross-functional model. In some cases, these AI teammates function similarly to lower-level managers, providing durable connective tissue across the organization rather than being confined to a single function or initiative. Examples include an AI project manager coordinating work across departments (requesting updates, tracking progress) or an AI knowledge curator continuously gathering artifacts to maintain a centralized repository.

