LLMs.txt directory

Spotter personal memory

Personal memory is what Spotter knows about you specifically — your role, your context, and your preferences. It is not about your data; it is about you as a user. Personal memory applies only to your sessions and does not affect how Spotter answers anyone else.

Every user builds personal memory through conversation. You teach Spotter by stating what you want it to know. Spotter does not learn from your query patterns passively — nothing is saved unless you tell it explicitly.

Before you start

Spotter saves personal context only when you state it clearly — for example, "Remember I’m Head of CS for EMEA" or "Always show me tables." Conversations without explicit instructions do not generate personal memory.

Shared memory versus personal memory

Shared memory Personal memory

---

---

---

Built by

Users with Manage Memory access (and from Liveboards)

Any user

Applies to

All users on that data model

You only

Covers

Business definitions, query patterns

Role, preferences, personal context

Managed via

Memory Sources page

Conversation

Example

"Revenue excludes returns"

"I’m Head of CS for EMEA, focused on renewals"

Two layers of personal memory

Personal memory has two distinct layers, applied at different times.

Layer 1 — Your profile

Who you are and how you want to interact with Spotter, regardless of which data model you are querying. Your profile comes with you in every session.

What you tell Spotter How it helps

---

---

"I’m Head of CS for EMEA, focused on renewals"

Frames every response through a CS leadership lens

"I manage a team of 6 CSMs"

Scopes team-level queries to your team correctly

"I prefer tables over charts for trend data"

Returns tabular output by default

"Keep answers concise — I don’t need long explanations"

Leads with the answer, skips preamble

"Always show me month-over-month, not absolute numbers"

Applies MoM framing without you specifying each time

"Respond in French"

Responds in your preferred language

Layer 2 — Data model-specific personal context

Your personal preferences, scoping, and definitions for a specific data model. This layer applie only when you are querying that model.

What you tell Spotter How it helps

---

---

"When I ask about pipeline, I mean EMEA only" (GTM model)

Filters pipeline queries to EMEA whenever you are on GTM

"For me, performance means revenue attainment vs. quota" (GTM model)

Uses your definition, not the default one

"Churn for me means logo churn, enterprise accounts only" (CS model)

Applies your definition when you are querying CS data

"My accounts are my 30 EMEA enterprise accounts" (CS model)

Filters correctly whenever you say "my accounts"

Personal terms can differ across data models. Spotter keeps them separate and applies the right definition based on which Model you are querying.

Term Your definition on GTM model Your definition on CS model

---

---

---

"Performance"

Revenue attainment vs. quota

Renewal rate and NPS

"Churn"

—

Logo churn, enterprise accounts only

"Pipeline"

EMEA open opportunities only

—

How Spotter fetches personal memory:

Querying the GTM data model?
  ✓ Your profile           — always fetched
  ✓ Your GTM preferences   — fetched because GTM is in scope
  ✗ Your CS preferences    — not fetched (not in scope)

There is no fixed list of what is worth teaching. If telling Spotter something about yourself would change how you want it to answer, it is worth saving.

How to teach Spotter

Teach Spotter through conversation — state what you want it to know, and Spotter saves it with a confirmation.

What triggers a save:

  • "Remember I’m Head of CS for EMEA focused on renewals"

  • "Always show me tables, not charts"

  • "My renewal target is 95%, not the team’s 85%"

  • "When I say 'my accounts', filter to my EMEA book"

Vague statements do not trigger a save. "That looked off" or "I want better answers" are not saved — Spotter requires explicit intent.

What Spotter confirms:

When personal memory is saved, Spotter confirms inline:

User: "Always show me results in a table. And remember my region is EMEA."

Spotter: "Got it — I’ve saved two things to your personal memory: output format set to table, and your region set to EMEA."

Teach once, applied in every session after:

Session 1 User: "Remember I’m Head of CS for EMEA. My focus is renewals and adoption." Spotter: "Got it — saved to your personal memory: role is Head of CS for EMEA, focus areas are renewals and adoption."

Session 2 (new conversation, no re-explanation needed) User: "How are my accounts doing this quarter?" Spotter: (applies EMEA region filter, leads with renewal risk and adoption gaps) "Here’s the picture for your EMEA book this quarter — three accounts are below your renewal threshold and flagged at risk…​"

Priority order

When Spotter forms an answer, it weighs multiple layers of context. Personal memory sits in the middle — below hard org rules, above the general data model context.

Explicit in-query instruction    ("show me absolute numbers this time")
        ↓ overrides
Spotter instructions             (Org-wide hard rules — always win)
        ↓
Personal memory                  (your context frames the answer)
        ↓
Shared memory                    (data model context, all users)

Hard guardrails in Spotter instructions always win. No personal preference can override a compliance rule or topic deflection your org admin has configured. For softer preferences — format, tone, scope — Spotter applies your personal memory when instructions do not specify otherwise.

Note: Behavior for soft preference conflicts between Spotter instructions and personal memory is not guaranteed to be consistent in this release.

What does Spotter know about you?

Ask Spotter directly:

"What do you know about me?"

Spotter summarises all saved personal memory for your account.

To understand why a specific response was personalised:

"Why did you filter to EMEA?"

Spotter explains which saved context it applied for that answer.

Correcting and deleting personal memory

Both happen through conversation.

To correct: State the updated information. Spotter overrides the previous entry and confirms.

"My region is now DACH, not EMEA."

To delete a specific memory:

"Forget that my region is EMEA." "Delete my output format preference."

To clear all personal memory: Available from user settings.

Enabling personal memory

For admins

Personal memory learning is off by default in Early Access. Two steps to enable it for your org.

Step 1 — Enable memory (if not already on):

  1. Navigate to cluster-level admin settings.

  2. Toggle on "Memory from Liveboards and conversations".

Step 2 — Enable personal memory for all users:

  1. Navigate to the same admin settings tab.

  2. Toggle on "Enable personal memory learning for all users".

When this second flag is off, end users cannot generate or see personal memories. Users with Manage Memory access can still write shared data model memory from conversation.

For TSE deployments, the personal memory flag is available at the Org level — TSE Org admins can control personal memory independently for their Org.

What happens when the flag is turned off:

  • No new personal memories can be written by end users.

  • Existing personal memories are preserved — they are not deleted.

  • Re-enabling the flag makes all existing personal memories active again immediately.

For users

When personal memory is enabled for your Org, you control your own participation from your profile settings.

Navigate to Profile settings > Personal Memory toggle.

  • On (default when Org flag is enabled): Spotter saves personal context from your conversations.

  • Off: Your conversations do not generate personal memories. Existing personal memories are preserved but not applied.

Early Access limitations

  • No memory management UI. View, correct, and delete personal memories through conversation only. A dedicated personal memory view is planned for a future release.

  • Active learning only. Spotter saves personal context when you explicitly teach it. It does not infer preferences from your query patterns passively.

  • No admin visibility into personal memories. Admins cannot view the personal memories of individual users in this release.

  • No import. There is no mechanism to bulk-import personal context in this release.

Permissions summary

Action Who can do it

---

---

Enable the master memory flag

Admins only

Enable personal memory learning for all users

Admins only

Write personal memory from conversation

Any user (when Org admin flag is on)

View own personal memory

Any user — ask Spotter: "What do you know about me?"

Correct or delete own personal memory

Any user, via conversation

Turn personal memory off for themselves

Any user (Profile settings toggle)

View another user’s personal memory

Not available in this release