Skip to main content

Noah AI Agent

The Noah AI Agent is a conversational assistant that helps you analyze data, generate reports, troubleshoot issues, and optimize your AI systems through natural language queries.

Agent Capabilities​

Architecture​

Using Noah AI​

The agent is available through multiple platforms for convenient access wherever your team works.

Available Platforms:

  • Slack Direct Messages
  • Slack Channels (mention @Noah)
  • Microsoft Teams
  • Web Dashboard Chat

Query Examples​

Project Analysis​

Simple Queries:

@Noah show me all projects
@Noah what's the status of Customer Support Chatbot?
@Noah list projects with active alerts
@Noah which projects have high drift scores?

Advanced Queries:

@Noah compare drift scores for Customer Support Chatbot 
between last week and this week

@Noah show me projects where cost increased by more than 20% 
in the last 30 days

@Noah find all runs with PII detections in the last 7 days

@Noah what's the average latency for gpt-4 vs gpt-3.5-turbo 
across all projects?

Metric Analysis​

@Noah show cost breakdown by model for December

@Noah what's causing the latency spike in Product Recommendations?

@Noah analyze drift trend for Customer Support over last quarter

@Noah compare readability scores before and after prompt change

@Noah show me all requests that exceeded $0.50 cost

Alert Management​

@Noah list all critical alerts

@Noah show alert history for last 48 hours

@Noah what alerts were triggered most frequently this month?

@Noah summarize alert resolution times by severity

@Noah show me alerts that took longer than 2 hours to resolve

Report Generation​

@Noah generate monthly report for Customer Support Chatbot

@Noah create executive summary for all projects

@Noah prepare compliance report for Q1 2024

@Noah generate cost optimization recommendations

@Noah create performance comparison report for all models

Troubleshooting Assistance​

@Noah why is my drift score increasing?

@Noah help me understand why run #456 failed

@Noah what's causing high costs in Product Recommendations?

@Noah debug latency issues in Content Generator

@Noah why are my alerts not triggering?

Conversational Context​

Noah remembers conversation context, enabling natural follow-up questions without repeating information.

Example Conversation Flow:

Conversation Example:

User: Show me projects with high latency

Noah: Here are 3 projects with latency >5s:
1. Product Recommendations (avg: 6.7s)
2. Content Generator (avg: 8.2s)
3. Translation Service (avg: 5.4s)

User: What about the first one?

Noah: [Understands "first one" = Product Recommendations]

PRODUCT RECOMMENDATIONS - LATENCY ANALYSIS

Current Metrics:
• Avg Latency: 6.7s (target: <3s)
• P95 Latency: 12.3s
• Timeout Rate: 2.3%

Contributing Factors:
1. Large Context Windows (avg 3,200 tokens)
2. Complex Prompts (multiple retrieval steps)
3. Model Selection (gpt-4 slower than gpt-4-turbo)

User: Try the third recommendation

Noah: I can help you switch to gpt-4-turbo. This will:
• Reduce latency by ~30% (from 6.7s to ~4.7s)
• Maintain similar quality
• Reduce cost by ~50%

Report Output Format​

When generating reports, Noah provides structured summaries with key metrics, identified issues, and actionable recommendations.

Sample Report Structure:

SectionContent
OverviewHigh-level summary of status
Key MetricsSuccess rate, latency, cost, drift
AlertsCritical issues and resolution status
Top IssuesMost impactful problems
RecommendationsPrioritized action items
AttachmentsFull PDF report with details

Best Practices​

Effective Queries:

  • Be specific about time ranges and project names
  • Use comparison queries to identify trends
  • Ask for explanations when metrics change
  • Request recommendations for optimization

Report Scheduling:

  • Set up automatic weekly/monthly reports
  • Include relevant stakeholders in distribution
  • Customize report content for audience
  • Archive reports for historical reference