curate-me.ai in Action
This blog runs on curate-me.ai. Everything below is live data from the platform — real-time metrics, request tracing, cost trends, and the governance chain that protects every LLM call.
Real-Time Metrics
Live metrics from the curate-me.ai gateway. Cost, request volume, latency, and fleet status update every 30 seconds. Full breakdown available on the cost dashboard.
Request Trace — Governance Chain
Every LLM request passes through a 5-step governance chain before reaching the provider. Rate limiting, cost estimation, PII scanning (blocks passwords, credit cards, personal data), model allowlists, and human-in-the-loop approval gates. Full traces on the tracing dashboard.
Cost Trend (7 Days)
Compliance Pipeline Status
All 5 governance checks are active. Each step short-circuits on denial — blocked requests never reach the LLM provider.
Live Cost Tracking
Every LLM call is tracked in real time. Budget caps prevent runaway spending. Cost breakdown by model and agent available on the cost dashboard.
Agent Fleet
10 specialized agents running in secure managed containers. Each has defined capabilities, an AI model, schedule, and cost budget. Managed via the fleet dashboard.
Activity Feed
Real-time agent event stream. Full traces available on the activity dashboard.
Knowledge Base
Knowledge stored locally for fast reads and synced to the cloud for cross-agent access. Every blog post, research finding, and agent discovery is indexed. Searchable via chat or the KB dashboard.
3-Tier Memory System
Most AI agents are stateless — every conversation starts from scratch. curate-me.ai implements a 3-tier memory architecture that lets agents remember who you are, how you behave, and what you're doing right now.
Profile Memory
MongoDB · PermanentLong-term preferences and attributes: account settings, domain expertise, budget ranges, provider preferences. Persists indefinitely.
Pattern Memory
MongoDB · DecayingBehavioral insights learned automatically: interaction patterns, decision tendencies, preference evolution. Decays over extended periods.
Session Memory
Redis · 30min TTLEphemeral context for the current conversation: message history, viewed items, current intent. Expires after 30 minutes of inactivity.
How Memory Flows Into Agent Execution
Pattern memory updates automatically after significant interactions. No user configuration required.
Fleet Templates
Development Team
Code review, CI, docs, and feature development with human approval.
Customer Support
Support triage, escalation handling, and analytics for customer service teams.
Personal Productivity
Inbox manager, calendar optimizer, and research assistant.
Hospital CFO
Financial orchestrator with specialist agents for healthcare revenue cycle.
Dashboard Capabilities
Agent Learning
Agents don't just execute — they learn. Every experiment, digest, and task writes insights to the shared knowledge store. Future runs build on past successes and avoid repeated failures.
Immutable Audit Trail
Immutable Audit Trail
patent-pendingPer-org SHA-256 hash chain with EU AI Act risk auto-classification. Every AI decision is tamper-proof.
chat/completions
chat/completions (PII detected)
HITL: approved by admin@org.com
chat/completions
chat/completions (security scan triggered)
EU AI Act Classification
Auto-classified based on model tier, cost, PII presence, and governance action. Full applicability: August 2, 2026.
Full Observability Dashboard
Everything on this page is a preview. The full dashboard includes distributed tracing, flamegraphs, per-model cost breakdown, session replay, and compliance scoring.
Technical Deep Dive
Architecture diagrams, API reference, governance config, and OpenClaw integration details.
Ready to Deploy?
Everything on this page is powered by curate-me.ai. Deploy your own agent fleet with governance, cost tracking, and full observability.