Problem Definition:

Rovo Chat does not currently support user-level instructions — persistent personal preferences that shape how Rovo interacts with an individual user across all conversations. Every Rovo conversation starts from zero context, requiring users to repeatedly provide the same background information, correct the same assumptions, and re-establish the same preferences in every new chat.

This is particularly impactful for power users and administrators who interact with Rovo frequently and have consistent context that applies to every interaction. In large organisations (e.g., 20,000+ users across diverse roles, teams, and technical competencies), this leads to significant repetitive prompting and slower time-to-value from Rovo.

Other AI chat products (e.g., Claude, ChatGPT) address this through a user profile or custom instructions configuration where users define standing directives that are applied to every interaction.

Note: This is distinct from Rovo Memory, which stores facts recalled from conversations. This request is for a deliberate, structured settings surface where users proactively define persistent instructions that control how Rovo responds, not what it remembers.

Example Use Cases:

• "I'm a Jira administrator. Default to admin-level explanations rather than end-user guidance."

• "Always respond in Australian English."

• "When I ask about Confluence, I'm referring to our Cloud environment, not Data Center."

• "Keep responses concise. Don't repeat my question back to me."

• "I work in the Platform Engineering team. Our Jira spaces are PLAT, INFRA, and DEVOPS."

Suggested Solution:

A user-controlled preferences section accessible from the user's profile or Rovo Chat settings, with the following characteristics:

1. Free-text input where users can define standing instructions (role, communication preferences, formatting, team context, etc.)
2. Applied to all Rovo Chat interactions (excludes agent-specific conversations where agent instructions take precedence)
3. Cannot override site-level instructions — organisational policy and guardrails remain enforced
4. Editable at any time, with changes applying to new conversations

This would form the middle tier of a three-tier instruction hierarchy:

This hierarchy allows organisational policy to be enforced consistently while giving individual users the ability to personalise their experience without requiring admin intervention or agent-level configuration for every possible user context.

Business Justification:

• Improves Rovo adoption by making it immediately useful from the first message of every conversation

• Reduces repetitive prompting across large user bases

• Empowers users to self-serve personalisation without admin involvement

• Aligns Rovo with user expectations set by competing AI chat products

• Particularly valuable for organisations with diverse roles, teams, and technical competencies where a one-size-fits-all response style is insufficient

Workaround:

N/A