DESIGN PRINCIPLES · COMEX VISION 2027
Six principles guide every interface.
From design to deployment. The platform constitution that grounds them →
Intent first, not navigation.
The user expresses what they want to know. The interface understands the intent and shortens the path. No deep menus. No wizards with 18 filters.
A CMI manager says "show me the ‘On Board’ pain points this week" and the interface configures itself. No filters to toggle one by one.
See Studio →AI investigates, humans decide.
AI does the analytical work: segmenting, comparing, identifying anomalies. Humans keep their hand on interpretation and the final decision.
Reviews doesn't show you a table of 10,000 reviews. It shows "Texture is your serum's weak spot — 3 key verbatims, 2 competitors winning, here's why."
See Reviews →Artifacts, not answers.
Each interaction produces a re-usable object: chart, scorecard, brief, insight card. The output enriches the workspace, not just the conversation.
When Echo answers "the Gen Z expectations on clean skincare cluster on 3 aspects", it generates a downloadable Insight Plan with findings, evidence, recommendations. Not just text in a chat.
See Echo →Proactive, not reactive.
The interface doesn't wait to be questioned. It watches, detects, and proposes.
Pulse delivers a personalized weekly briefing. Trends alerts when a signal crosses a threshold. Reviews flags when a competitor gains on a key aspect — before the next board meeting.
See Pulse, Trends →Transparent reasoning.
AI shows its work. Every insight displays its sources, method, and confidence level. The user can drill down into the reasoning. Mandatory: EU AI Act effective August 2026 — transparency is a legal obligation, not a nice-to-have.
An auto-generated narrative in Pulse displays [source: Q1 survey, n=2340, confidence: high]. The user clicks to see the underlying verbatims and the model that produced the synthesis.
See Method →Progressive autonomy.
The interface starts by assisting (AI suggests), then proposes to act (AI executes and summarizes) once trust is established. The user controls the level of delegation.
Echo starts by answering questions. Then it proposes "Want me to prepare the Q2 competitive brief automatically every Monday?" The user chooses their comfort level.
See Agents →