
AI Agent Skill Difficulty Points Explained
Understand Skill Difficulty Points for AI agents in hospitality. Learn how to combine skills, manage capacity, and grow your assistant with your operations.
AI Agents as Team Members: The Mental Model
Every assistant on inHotel has potential.
Just like a person in your team, an assistant can learn new capabilities, take on more responsibility, and grow over time. But there is always a limit to how much someone can effectively handle at once.
That’s where Skill Difficulty Points come in.
Skill Difficulty Points measure how difficult a skill is to acquire and perform, while your plan determines how many of those points an agent can actively carry at once.
In the sections below, you’ll see what Skill Difficulty Points represent, how they work in practice, and how you can use them to shape your assistant’s capabilities.
What Are Skill Difficulty Points?
Just like roles and responsibilities vary in difficulty for people, skills also vary in how demanding they are to learn and perform.
Each skill has a difficulty level that reflects how hard it is for an AI agent to acquire and apply reliably. Simple skills usually involve focused tasks with clear instructions, while more difficult skills require deeper reasoning, structured workflows, or expert judgment.
The more difficult a skill is to learn and perform well, the more Skill Difficulty Points it requires.
How Skill Difficulty Points Are Evaluated
Skill Difficulty Points estimate how hard it is for an AI agent to acquire, apply, and reliably perform a skill in real operating conditions.
In other words, they reflect the difficulty of teaching an agent a capability, not just the value of that capability. Some skills are easy to learn and perform consistently. Others require more reasoning, more structure, more judgment, or deeper expertise.
A skill becomes more difficult as it requires more of the following:
- contextual understanding
- reasoning depth
- multiple steps or procedures
- exceptions and decision paths
- ambiguity handling
- tool or system interaction
- specialized domain expertise
- consistent performance in real situations
This is why two skills that sound equally useful may carry different point values. The score depends on how difficult the capability is to acquire and execute well.
The 1–5 Skill Difficulty Rubric
To make Skill Difficulty Points easier to understand, we group skills into five levels.
🟢 1 point → Retrieve (Minimal reasoning required)
Easy to acquire. The agent handles a simple, single-step task with little ambiguity or judgment.
Example: answering standard guest questions such as check-in time, breakfast hours, or Wi-Fi availability.
🟡 2 points → Interpret (Requires understanding, not planning)
Still relatively easy to acquire. The agent must understand and apply context, but there is no fixed multi-step process.
Example: understanding a guest’s preference for “something casual, local, and vegetarian-friendly” and suggesting a suitable nearby dining option.
🟠 3 points → Execute (Requires following a known, predefined process)
Moderately difficult to acquire. The agent follows a clear, checklist-style procedure. Branching is predefined.
Example: handling a housekeeping request by confirming details, classifying the request, and logging it correctly.
🔴 4 points → Decide (Requires choosing between valid paths using judgment)
Difficult to acquire. The agent works through a defined problem with rules, exceptions, and multiple valid paths. The goal is clear, even if decisions are not straightforward.
Example: analyzing a hotel’s P&L to identify material variances, distinguish likely operational causes from one-off anomalies, and prepare a management-ready summary.
🟣 5 points → Define & solve (Requires defining the problem before solving it)
Highly difficult to acquire. The agent must decide how to approach an open-ended problem, determine what matters, and adapt as it goes.
Example: investigating why profitability is declining, deciding which data to examine, identifying likely causes, and drafting management recommendations.
As a simple rule of thumb, lower-point skills are easier to teach and more focused in scope, while higher-point skills require more autonomy, stronger judgment, and greater ability to determine the right path to success, progressing from following known paths to choosing between them and ultimately defining them.
Skill Difficulty Points and Hosting Costs
Skill Difficulty Points also serve as a reference for the minimum hosting and support cost required to operate a skill reliably. As skills become more difficult, they typically require:
- more advanced language models
- higher token usage
- more orchestration and system resources
To reflect this, each difficulty level is associated with an indicative monthly cost:
- 🟢 1 point → Retrieve = from €5/month
- 🟡 2 points → Interpret = from €10/month
- 🟠 3 points → Execute = from €15/month
- 🔴 4 points → Decide = from €20/month
- 🟣 5 points → Define & solve = from €50/month
These values are reference points. Actual hosting and support costs may vary depending on factors such as model selection, usage volume, workflow complexity, actual cost.
How AI Agent Skill Difficulty Points Work
Now that you know what Skill Difficulty Points are, here’s how to use them.
Each AI Assistant or Persona (any agent) has a Skill Difficulty Point allowance based on your plan. You can combine skills freely, as long as the total difficulty stays within that allowance.
For example, you might equip your assistant with several simple skills, combine a more advanced workflow with a few supporting capabilities, or build a layered setup with multiple workflows supported by smaller skills. In practice, this works much like assigning a mix of responsibilities to a team member.
Browse the Skills Catalog to see what your assistant can learn next.
You can also view each skill’s difficulty directly in your Customer Portal as you explore available options.
Plans, Capacity, and Growth of AI Agents
Each plan defines how much your assistant can grow, and how much complexity it can handle at once.
Smaller plans are ideal to explore and run a focused setup, much like onboarding a new team member with a limited scope. As plans grow, they allow you to combine more skills and handle more complex operations, similar to experienced employees who can context-switch across different types of work throughout the day.
At the highest tier, limits are removed entirely, like having a top-performing team member who can handle almost anything you throw at them.
As your needs grow, your assistant can grow with you.
Skill Ownership vs Usage Limits
Importantly, ownership and usage are separate concepts.
You can acquire as many skills as you like and keep them in your library. Skill Difficulty Points only limit what your assistant can actively use at a given moment.
This gives you flexibility to prepare for seasonal operations, test new workflows, and switch setups without losing anything.
Skill Assignment per AI Agent
Skills are assigned per assistant to keep each role focused and reliable.
Each AI Assistant or Persona acquires skills individually. For example, if both your Finance Assistant and General Manager Assistant need to analyze P&L statements, each assistant must be equipped with that skill.
We plan to introduce skill transfer within the same organization in the future, so you will be able to reuse capabilities more easily across assistants.
Ongoing Management and Flexibility
Skill Difficulty Points are designed to support ongoing management and evolution, similar to how you would develop a human team over time.
As your operations change, you can refine your assistant’s setup by activating, deactivating, or reshaping skills to match current priorities. This allows your assistant to stay aligned with your business, much like a team member whose responsibilities evolve over time.
Summary: AI Agent Skill Difficulty Points
Skill Difficulty Points define how much your assistant can handle, similar to a person’s capacity for work. Each skill carries a difficulty (1–5 points). You can combine skills within your allowance, own more skills than you actively use and switch them as needed, and upgrade to increase your assistant’s overall capacity.
Your assistant grows with your business, one skill at a time.
What are AI agent Skill Difficulty Points?
AI agent Skill Difficulty Points define how much complexity an AI assistant can handle at any given time. Each skill is assigned a difficulty score based on how demanding it is to learn and execute. Simple tasks like answering questions may require 1–2 points, while multi-step workflows or automations require 3–5 points. An assistant has a limited number of points available depending on its plan, and all active skills must fit within that allowance. This system helps structure how capabilities are combined, ensuring the assistant remains effective and manageable rather than overloaded.
How many skills can an AI agent have at the same time?
The number of skills an AI agent can use simultaneously depends on its total Skill Difficulty Point allowance. Instead of counting skills individually, inHotel measures their combined complexity. For example, an assistant could run several simple skills or a smaller number of more advanced workflows, as long as the total stays within the limit. This approach provides flexibility, allowing operators to prioritize either breadth or depth depending on their operational needs.
Can I own more AI agent skills than I actively use?
Yes, you can acquire and store more skills than your assistant can actively use at once. Skill Difficulty Points only limit which skills are active at a given moment. This means you can build a library of capabilities and switch between them as needed. For example, you might activate certain skills during peak season and replace them later with others more relevant to different operational priorities. This flexibility allows continuous experimentation and adaptation without losing access to previously acquired capabilities.
What determines the difficulty level of an AI agent skill?
A skill’s difficulty level is based on how hard it is for an AI agent to acquire, apply, and perform reliably in real operating conditions. Lower-difficulty skills usually involve simple retrieval or contextual interpretation, while higher-difficulty skills require following structured procedures, making decisions within defined problems, or even determining the path to solve an open-ended problem. Difficulty reflects the level of understanding, reasoning, judgment, and operational structure the skill requires, not its price or business value. This creates a consistent way to combine different types of skills within an assistant’s capacity.
Do AI agent Skill Difficulty Points affect skill pricing?
Skill Difficulty Points are separate from pricing. They measure the difficulty of acquiring and reliably performing a skill, as well as the capacity it requires, not its price. A simple skill may be free or paid, and the same applies to advanced workflows, with pricing determined independently by the skill publisher. However, difficulty points can indirectly influence pricing, as they indicate the minimum hosting and support cost required to operate a skill, which publishers may factor into their pricing decisions. This separation ensures that the system remains fair and flexible, allowing users to choose skills based on their operational needs rather than being constrained by how difficulty and pricing interact.
Can one AI agent skill be used across multiple roles or assistants?
Currently, each AI assistant or persona must be equipped with its own set of skills. If multiple roles, such as a Finance Assistant and a GM Assistant, need the same capability, each must have that skill assigned individually. This ensures that each assistant remains focused and reliable for its specific function. We expect to evolve to allow transfer of skills across roles within the same organization in the future.
How do Skill Difficulty Points scale as my business grows?
As your business grows, higher-tier plans provide more Skill Difficulty Points, allowing your AI agents to handle more complex and layered operations. Smaller plans are suitable for testing and focused use cases, while larger plans enable combining many workflows and capabilities. At the highest level, limits can be removed entirely, allowing unrestricted scaling. This mirrors how businesses invest in more capable team members as operations expand and become more sophisticated.
How should hotels choose which AI agent skills to activate?
Hotels should prioritize skills based on current operational needs and impact. Start with essential tasks such as guest communication or internal knowledge access, then expand into workflows that automate repetitive or complex processes. Because capacity is limited, it is important to balance simple and advanced skills strategically. Over time, teams can refine their setup by activating and deactivating skills as priorities shift, ensuring the assistant remains aligned with evolving business needs.