Answer extracted from the Unprompted: Real AI in IC podcast — listen to the full episode below.
How does an emotionally intelligent communications checker agent work—and what are its real limits?
Rebecca Lindsay built an agent that scores written communications against emotional intelligence principles using a defined rubric, but it only analyzes text and has no access to human context—missing critical nuance that matters to real conversations.
A rubric-driven approach to emotional tone
The agent works by evaluating written communications—emails, messages, announcements—against key emotional intelligence principles. Rebecca built it to give instant feedback on tone and approach, mimicking what an emotionally aware reviewer would flag.
The system uses a clear scoring rubric to evaluate each message. This keeps the feedback consistent and reproducible, though it also locks the agent into a fixed framework.
The blind spot: context and nuance
The agent's biggest limitation is that it cannot see the human context. An emotionally intelligent response depends heavily on who wrote the message, why, for whom, and what happened before—none of which the agent can access.
As a result, it flags patterns based purely on words and tone, missing the real story. A sharp response might be exactly right if the recipient has been asking the same question five times; it might be tone-deaf if delivered to someone in crisis. The agent can't tell the difference.
"We want to be giving you the tools to do great comms yourself, and that is precisely what AI can enable us to do at scale."
Rebecca Lindsay — Internal Comms Manager, Hyper Exponential. Rebecca leads AI experimentation across internal communications at a company where the CEO explicitly champions AI adoption across all roles. She has built multiple agents—tone of voice checkers, annual leave assistants, and emotional intelligence reviewers—as part of the company's AI enablement squad, turning organizational friction points into scaled solutions.
The unexpected value: reactions reveal what matters
Despite its limitations, the agent has already sparked something valuable: it triggers real conversations when people receive unexpected scores. When someone gets a low emotional intelligence rating they weren't expecting, they push back—and those conversations surface disagreements about what "good" communication actually looks like.
That friction is information. It reveals where people's intuitions differ from the rubric, and where the definition of emotionally intelligent communication is actually contested within the organization. Rebecca describes this in the episode as one of the most valuable outcomes—not the scores themselves, but the conversations those scores unlock.
See also
What prompt engineering techniques help ensure better AI agent outcomes?
Rebecca recommends asking the AI tool to clarify your requirements before running the prompt, since you don't know what you don't know. This approach helps surface hidden assumptions and improves the quality of the agent's output.
What are the key steps for implementing and rolling out an AI agent in an organization?
The three steps are: first, identifying the need through data or observed friction; second, creating the agent itself; third, behaviorally designing the rollout to ensure adoption and understanding across the organization.
How can internal communicators identify which problems are suitable for AI agent solutions?
Rebecca identifies needs through data and evidence, such as from internal comms surveys. For example, the annual leave assistant was created after survey data revealed time poverty among employees.