AI summary automatically creates short, clear overviews of your observations in your preferred language. Instead of reading through full observation details, you get instant insight into what happened at a glance.
It is an optional feature that's not enabled by default for all organisations. To activate AI summaries, please contact our support team.
🚧 The AI summary is a new feature currently in active development, so certain aspects may change over time. We welcome your feedback and encourage you to share your thoughts with us via email.
Where you'll see AI summaries
AI summaries enhance observation visibility across the entire Falcony platform:
Observation list page: Each observation card includes its AI-generated summary according to the user locale setting
Observation table page: Summaries appear in table view for efficient bulk review and allows to show summaries in all available languages
We also plan to show AI summaries on following pages:
Home page: Observation cards display summaries for quick overview of recent activity
Audit page: Observation cards show relevant summaries during audit processes
How AI summary works
When you submit an observation, we generate a text prompt for the AI language model using the information from your observation along with specific instructions for creating high-quality summaries. Within seconds, it generates concise summaries in your organisation's configured languages.
For more technical details about our AI infrastructure, read more in our AI processing: Privacy and Security help article.
What information is used by AI
Information available to AI for summary generation:
Observation description and details
Organisation name
Reporter name and job title (except special cases mentioned below)
Observation template
Questions and their help text (except special cases mentioned below)
Answers to the included questions
Information excluded from the analysis:
Hidden reporter and "External" users reporting through the open links
Hidden answers and answers with limited lifetime
Answers to follow-up questions
Comments
Attachments (this feature may be added in the future)
Observation activity history
🔒 Privacy protection: We filter out information with restricted visibility, so AI only processes data that all users with observation viewing rights can already access. This is done to prevent the AI summary from leaking sensitive answers to users that shouldn't have right to see that information.
Limitations
Personal Identifiable Information (PII): If PII appears in any non-hidden field, it may appear in the AI summary. To prevent this, ensure sensitive information is captured in appropriately configured hidden answers.
Whistleblowing: AI summaries feature cannot be used in whistleblowing organisations.
When AI summaries are generated
Actions that trigger summarisation:
Creating a new observation generates initial AI summary
Editing an existing observation replaces previous AI summary
Actions that do NOT trigger summarisation update:
Editing question's name or help text
Changing user's name or job title
Adding comments to observations
👍 Rule of thumb: Every time you click the edit pencil icon on the observation page and submit your changes, it will trigger a new summarisation. Actions performed elsewhere that affect multiple observations will not automatically trigger summarisation. If you need to update a large number of AI summaries, please contact our support team.
Toy example
Step 1: Sarah Chen, Front Desk Supervisor, submits an observation in the "Maintenance Request" category with the following information:
Step 2: The language model receives the data and a set of configured languages:
Sarah Chen, Front Desk Supervisor reported the Maintanance Request observation:
Description: Guest in 404 called this morning complaining about...
Priority level: Medium
Guest details: [Redacted]
Write a clear and concise summary in Finnish, Swedish and English
Step 3: Users see the summary according to their locale settings. For example, a Finnish user would see the Finnish version of the summary in the observation card.
Multilingual support
Organizations can configure multiple languages for AI summarisation. Users will see summaries in their preferred language based on their locale settings. Contact our support team to configure languages for your organisation.
Frequently asked questions
Q: How can I ensure personal information is never included in AI summaries?
A: Use appropriately configured hidden fields for sensitive information. Any data marked as hidden will be automatically excluded from AI processing.Q: If someone reports an accident and personal ID information is quickly removed after reporting, could it still appear in the AI summary?
A: No. When you edit an observation to remove information (including personal ID data), this triggers the generation of a completely new AI summary. The existing summary is cleared and replaced with a fresh summary based only on the current content of the observation. Any information that was removed during editing will not be included in the new summary.
This means if personal information was accidentally included initially, removing it through observation editing ensures it won't appear in the AI summary going forward.Q: What happens to existing summaries when I edit an observation?
A: When you edit an observation, the existing AI summary is cleared and a completely new summary is generated based on the updated information.Q: Can I disable AI summaries for my organisation?
A: Yes, AI features can be controlled through organisational settings. Contact our support team to discuss configuration options for your specific needs.