AI Memory
Show Locari what it should know about your properties and applicants — and correct entries with one click before they influence the next response.
Locari stores in its AI memory what it has learned about your properties, listings, and applicants during each interaction. This prevents it from asking the same question twice and enables more precise responses. You can view the memory at any time, correct individual entries, or delete them — Locari only forgets what you tell it to forget.
What Locari does automatically: Store facts from conversations, clarifications, and your manual replies — organized by category (e.g. pet policies, parking situation, style preferences). What you decide: Which entries remain, which are corrected or deleted. Locari only changes its behavior after you save.
Common Tasks
- View an applicant's AI memory
- Correct a stored fact
- Delete an entry
- Teach Locari your preferences
- Track the learning history
- Reset all memory
Memory entries always apply to one specific property, listing, or applicant — not globally. For account-wide communication rules, use AI Communication.
How to
View AI Memory
- Open the applicant, property, or listing profile.
- Click AI Memory (brain icon) in the top right.
- The modal shows all stored facts, grouped by category — for example:
- Pet policies: "Small cats allowed, prior discussion required"
- Style preference: "The Müller family prefers short, factual emails"
What Locari stores: Details from clarifications, corrections from your manual replies, applicant preferences (e.g. desired move-in date of July 1st). The source of each entry — "Locari AI" or "manual" — is visible in the history. What you do: Check whether stored facts are still accurate.
Correct a Fact
- Open AI Memory → find the entry you want to change.
- Click the pencil icon — the value becomes editable.
- Enter the new value and click Save.
- Example: Correct the entry "Income: 2,400 €" to "Income: 2,800 € (salary increase confirmed)".
Locari uses the corrected value from the next interaction onwards.
Delete an Entry
- Open AI Memory → find the entry.
- Click the trash icon → confirm deletion in the confirmation dialog.
- Example: Remove the outdated entry "Viewing on May 3rd" after the appointment was cancelled.
Note: A deleted entry cannot be restored. Locari treats that area as unknown after deletion and will ask again if needed.
Teach Locari Preferences
Entries do not only come from clarifications — you can also add facts manually:
- Open AI Memory → click + Add fact.
- Select a category (e.g. Policies, Amenities, Location).
- Enter the key and value.
- Example: Category "Policies", key "Pets", value "Small dogs allowed — max. 10 kg".
How Locari learns from this: The new entry immediately becomes part of the knowledge base. The next time an applicant asks about pets, Locari responds precisely — without asking you first. What you decide: Which categories and values are stored. Locari never invents facts.
Views and Fields
Knowledge Base
Visible per stored entry:
- Category — e.g. amenities, location, policies, utilities, general
- Key — unique name of the fact (e.g.
pets,parking_space,move_in_date) - Value — the stored content; can be text, number, list, or yes/no
- Last updated — date and time of the last change
- Source — "Locari AI" (automatically learned) or "manual" (entered by you)
Learning History
Chronological list of all write operations:
- Timestamp of the change
- Category and key of the affected entry
- Operation: set, appended, or removed
- Source: Locari AI or manual intervention
The history grows up to 100 entries — older entries are automatically displaced once the limit is reached.
Entity Types
AI memory is available for these entities:
- Listing / Rental case — policies, preferences, clarification history
- Property — amenities, location, utilities
- Rental unit — unit-specific facts
- Applicant / Prospect — move-in wishes, household composition, style preferences
- Contact — general information about the contact person
Audit and History
The History tab in the AI Memory modal documents every write operation — when a fact was set, appended, or removed, and whether the source was "Locari AI" or a team member. The log helps you understand why Locari gave a specific response.
Retention: The history is retained as long as the entity (applicant, property, etc.) exists in the system. When you manually clear memory (the "Clear all" button), the knowledge base and history are immediately and irreversibly removed.
What the history shows: Every write operation with timestamp, category, key, and source — so you can understand why Locari gave a specific response in an email. What you do: Review the history and correct or delete any inconsistent entries.
Note: Locari does not forget automatically — unlike personalized AI in chat applications. Entries remain until you explicitly delete them.
Permissions and Multi-select
- Visible to: all team members of the owner account.
- Read entries: all members.
- Add, edit, delete entries: all members.
- Reset all memory ("Clear all" button): administrators only — the action is irreversible.
- Multiple entities at once: memory entries are managed per entity; no bulk editing. For communication rules that apply to all listings, use AI Communication — style rules, instructions, FAQ.