Imagine two tasks: Every morning, you want to see which guests are arriving and which rooms are not ready yet. You also want new guest requests reviewed so your team can prepare in time.
Both tasks use current data. Both can run regularly. Yet they call for different tools.
A report runs a saved analysis. An agent carries out an assignment, using AI on every run. That distinction helps you choose the right approach for each task.
What is a report?
A report presents data according to defined rules. You decide which information belongs together, how metrics are calculated and how the results should look.
Take an arrivals overview: reservation reference, arrival time, booked room category, assigned room and cleaning status. Reservations without an assigned room appear first. Guests who have already checked in appear in a separate section.
AI helps you build this analysis. Its definition is then saved. Every subsequent run retrieves current data and applies the same queries, joins, calculations and presentation rules, without a new AI run. The report documentation explains this in more detail.
Reports can read and analyse data. They cannot change data in your source systems or perform actions in those systems.
You save the report in Discontinue. Your team can open it there or have it delivered as a PDF, Excel or CSV file, for example. The file format is simply one way to receive the result.
What is an agent?
An agent receives an assignment that it carries out based on the current situation each time it runs. It uses AI and the connected systems you give it access to.
For example: Check new reservations for special requests. Summarise anything relevant to preparing the stay and notify the front office if something needs clarification.
This involves more than a fixed calculation. Messages written in different ways need to be understood in the context of the reservation.
An agent can also take actions, such as assigning a room or adding selected data to another system. Your instructions and configured permissions determine what it may do and when approval is required. See the agent documentation for more detail.
An agent does not have to change anything. An assignment that only checks information, interprets it and notifies someone when needed can also be a good fit.
The differences at a glance
| Question | Report | Agent |
|---|---|---|
| What repeats? | A defined analysis | An assignment carried out using current information |
| When is AI used? | During creation or AI-assisted edits; not when running the saved analysis | On every run |
| What does it produce? | A structured overview with tables, metrics or charts | An interpretation, message or completed action, depending on the assignment |
| Can it change data in connected systems? | No | Yes, where instructions and permissions allow |
| Can it use multiple sources? | Yes | Yes |
| When does it run? | On demand or on an enabled schedule | Manually or through separately enabled triggers, such as a schedule or supported event |
When should you use a report?
A report fits when you want to answer the same question regularly using the same rules.
For the morning shift, that might mean: Who is arriving today? Which reservations still need a room? Which assigned rooms are not clean yet? The data changes every day. The analysis stays the same.
This also applies to more extensive analysis. A report can combine data from multiple connected sources, compare periods and calculate metrics. You could, for example, combine guest request counts from a service system with occupied room nights from Apaleo. A metric such as “requests per 100 occupied room nights” makes periods with different occupancy levels easier to compare.
Defined thresholds, flagged exceptions and calculated correlations can also be part of a report. Multiple sources and sophisticated calculations do not automatically turn an analysis into an agent task. What matters is whether the logic can be defined in advance.
A report is particularly useful for:
- daily overviews for a shift or team;
- recurring comparisons between periods or locations;
- metrics with consistent definitions;
- lists that highlight missing information or exceptions using fixed rules.
When should you use an agent?
An agent fits when each run requires fresh interpretation or includes actions in connected systems.
Understanding guest requests: “We arrive after midnight and are travelling with a toddler” may contain information relevant to several parts of preparing the stay. An agent can review messages like this in the context of the reservation and flag open questions to the right people.
Assigning rooms: An agent can look for a suitable room for a new reservation and assign it within your rules and permissions. You define the criteria and which actions need approval.
Moving data between systems: Your team might maintain an existing handover list in Google Sheets or Excel. An agent can add or update selected data from Apaleo there, provided the connected systems and permissions support those actions. That is a data transfer assignment. If you need a recurring overview, you can create it directly as a report in Discontinue.
Notifications also deserve a closer look: If you simply want the same analysis delivered every morning, a scheduled report is enough. If new messages need to be interpreted to produce a relevant notification, an agent fits the task.
How reports and agents work together
You do not have to give the entire workflow to one tool. An agent can use a report as a data source.
For arrivals preparation, that could look like this:
- The report provides the overview. It shows today's arrivals, room assignments and cleaning status using fixed rules.
- The agent reviews the additional context. It reads relevant guest requests and identifies questions the team still needs to resolve.
- The agent carries out the assigned next step. It might notify the front office or make an explicitly permitted change. Before making a change, it checks the current state in the source system.
This keeps your metrics and tables consistently defined. Additional interpretation and any actions belong to the agent's assignment.
How to get started
Start with the task in chat. Ask to see the information, follow up and refine the result. Then ask yourself three questions:
- Do I need this analysis regularly with current data? Create a report.
- Does each run need to interpret new content or perform actions? Create an agent.
- Do I need a consistent overview followed by further work? Combine both.
You can build a report or agent from the context you have already developed in chat. Open a built agent from the chat menu in split view to edit and run it directly. Continue refining it in the same conversation.
Enable automatic triggers and schedules separately when everything is ready. Creating a report or agent alone does not start automatic operation.
Pick a task from your working day: What information should be available regularly, and what should happen next? That will tell you whether you need a report, an agent or both.