Skill (AI)
A discrete, reusable set of instructions that defines how an AI agent should approach and complete a specific task or category of work. Skills give an AI agent a defined playbook to follow, specifying the steps, tools, data sources, and output format relevant to that task. Rather than relying on a single general-purpose prompt, a well-designed AI system assembles the appropriate skill for each task on demand. In commercial real estate, skills might cover workflows such as underwriting a multifamily acquisition, abstracting a lease, assessing market risk for a submarket, or sizing a construction loan, with each skill encoding the methodology an experienced analyst would apply to that specific type of work.
Putting Skill (AI) in Context
An asset management team building an AI-assisted reporting workflow creates a dedicated skill for quarterly investor reporting that specifies which data sources to pull from, how to calculate key performance metrics, and what format the output should follow. When a team member triggers the reporting workflow, the AI retrieves trailing twelve months financials from the property management system, applies the firm’s standard variance analysis methodology, and produces a draft narrative, because the skill encodes those steps explicitly rather than leaving the model to interpret the task from scratch each time.
Frequently Asked Questions about Skill (AI)
How is an AI skill different from a prompt?
A prompt is a single instruction given to an AI model for a one-time task, while a skill is a structured, reusable set of instructions designed to be retrieved and applied consistently across many instances of the same type of work. Skills typically specify not just what to do but how to do it, including which steps to follow, what data sources to consult, and what the output should look like. For CRE teams, the distinction matters because a skill produces repeatable, auditable results rather than outputs that vary based on how a question was phrased on a given day.
Who on a CRE team is responsible for building and maintaining AI skills?
In most CRE firms today, skills are built by a combination of the analysts or operators who understand the workflow deeply and whoever on the team has responsibility for AI tooling, whether that is an operations lead, a technology director, or an external consultant. The subject matter expert contributes the methodology and the edge cases, while the person building the skill translates that knowledge into structured instructions the AI can follow. Skills should be reviewed and updated periodically as market conditions, firm methodology, or available data sources change.
Can one AI skill cover multiple property types or deal structures?
It is technically possible to build a broad skill that handles multiple property types, but narrower skills tend to produce better results because the methodology for underwriting a multifamily acquisition differs meaningfully from underwriting a net lease industrial asset or a construction loan. A well-designed AI system selects the appropriate skill based on the task at hand, so building a library of focused, purpose-specific skills is generally more reliable than consolidating everything into a single general skill. The tradeoff is that maintaining a larger skill library requires more ongoing governance.
What happens when an AI agent encounters a task that no existing skill covers?
When no matching skill exists, the AI model falls back on its general training, which means the output may be less structured, less consistent with firm methodology, and less reliably accurate for that specific task. This is one of the more common failure modes in early AI deployments, where teams underestimate how many distinct task types their workflows actually contain. Identifying gaps in skill coverage through regular review is important for firms that want their AI systems to perform dependably across the full range of their work.
How do AI skills relate to the broader concept of an AI agent?
An AI agent is the system that receives a task, decides how to approach it, and executes the necessary steps, while skills are the discrete playbooks that agent draws on to do specific types of work. The relationship is similar to an analyst and a firm’s standard operating procedures: the analyst has judgment and the ability to act, but the procedures define how particular categories of work should be handled. A more capable agent with a well-developed skill library can handle a wider range of CRE workflows reliably, while an agent without well-defined skills tends to produce inconsistent results even on familiar task types.
Click here to get this CRE Glossary in an eBook (PDF) format.