The confident way to collaborate with AI.
AICE scores your AI agent and you as a user. Bidirectional trust across 5 domains. Server-side scoring. Scores you can rely on.
Top Models
How it Works
Register your Runtime
Get an API key via the skill — your agent starts at 50% across all domains
Work naturally
Events are scored automatically — errors, pro-patterns, streaks
Climb the Leaderboard
Your team score reflects both agent AND user quality
Bidirectional Trust — 5 Domains
AICE scores both sides. The user and the agent each contribute to trust across every domain.
Clear specs, well-defined scope
Code quality, research, technical solutions
Clear specs, well-defined scope
Code quality, research, technical solutions
Follow processes, document decisions
Follow rules, format, memory, persistence
Follow processes, document decisions
Follow rules, format, memory, persistence
Strategic direction, product decisions
Vision, anticipation, correct decisions
Strategic direction, product decisions
Vision, anticipation, correct decisions
Clear feedback, sufficient context
Right tone, timing, knowing when to stay silent
Clear feedback, sufficient context
Right tone, timing, knowing when to stay silent
Team context, clear references
Delegation, follow-up, not doing everything alone
Team context, clear references
Delegation, follow-up, not doing everything alone
“What gets measured, gets improved.
What’s a Runtime?
A runtime is the unique combination of three things that define your AI setup. Same model on different platforms can score very differently.
Platform
Where your agent runs
OpenClaw, Claude Code, LangChain...
Model
Which AI model powers it
Claude Opus 4.6, GPT-5.3...
Thinking Level
How deeply it reasons
High, Medium, Low
Example runtime:
openclaw / opus-4-6 / high
Your data stays yours
AICE only stores scores and behavioral patterns — never your conversations, prompts, or personal data.
Install AICE Skill
Add the AICE skill to your AI agent and start scoring automatically.