Hugo

Hugo is SaliencyLab's conversational agent. It helps you move through the existing workflow from one conversation rather than replacing the underlying project, Market, RoastIQ, or BuyerLens surfaces.

Use Hugo when the task crosses several parts of the workspace or when you want help translating evidence into a next step. Use the dedicated product surface when you need to inspect complete source evidence, manage settings, or compare detailed results.

Ask with context

Useful prompts identify the project, creative decision, platform, market, and desired outcome. For example:

Review the latest video in this project for TikTok UK and tell me which opening edit deserves testing first.

A reliable prompt usually contains:

• Object: the project, creative, result, or competitor set
• Context: platform, market, audience, and campaign role
• Decision: what you need to approve, change, compare, or investigate
• Output: the format you want, such as a prioritized edit list or summary

Example requests

• “Summarize the strongest and weakest evidence in this RoastIQ result without dropping the confidence labels.”
• “Compare these two variants and explain which differences are material to the verdict.”
• “Turn the Brand Impact finding into one edit hypothesis for the next version.”
• “Find relevant market patterns for short-form skincare ads in the selected market, then separate observations from modeled conclusions.”

Keep control of actions

Review the project, asset, and destination before confirming an action. If Hugo refers to a result or creates a handoff, open the underlying record and confirm that it belongs to the intended workspace and project.

Verify the handoff

When Hugo starts or summarizes work, open the underlying project and result to inspect the evidence directly. The conversation is an interface to the workflow, not a separate source of truth.

Do not share credentials, API keys, customer lists, or unnecessary personal data in a conversation. For methodology-sensitive claims, verify the evidence type and confidence before sharing the answer outside the workspace.