Hyperfold AI

Observability

Monitor agent performance, trace requests, and set up alerting.

Overview

Hyperfold provides comprehensive observability through metrics, logs, traces, and alerts. Monitor agent performance, track negotiations, and get alerted to issues before they impact revenue.

Observability data is stored in Google Cloud Operations Suite (formerly Stackdriver) and accessible via CLI, API, and web console.

Key Metrics

Track key performance indicators:

bash
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
# View agent metrics
$ hyperfold metrics agents
 
AGENT METRICS (24h)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
NEGOTIATOR: sales-bot-01
  Sessions:          1,247
  Conversions:       412 (33.0%)
  Revenue:           $67,840
  Avg Order Value:   $164.66
  Avg Discount:      12.3%
  Margin:            46.2%
 
  Response Time:     p50: 120ms  p95: 340ms  p99: 890ms
  Error Rate:        0.3%
  LLM Tokens:        2.4M tokens ($48.00)
 
FULFILLMENT: fulfillment-01
  Orders Processed:  389
  Shipments Created: 385
  Avg Processing:    4.2 minutes
  Carrier Split:     USPS 45%, FedEx 35%, UPS 20%
 
# Export metrics to JSON
$ hyperfold metrics agents --output=json --since=7d > metrics.json

Metric Categories

CategoryMetrics
BusinessRevenue, conversions, AOV, discount rate
PerformanceResponse time, throughput, error rate
CostLLM tokens, API calls, infrastructure
AgentSessions, negotiations, bundle uptake

Dashboards

bash
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
# Open metrics dashboard
$ hyperfold dashboard open
 
> Opening https://console.hyperfold.io/dashboard...
 
# Available dashboards:
# - Agent Performance
# - Revenue Analytics
# - Negotiation Insights
# - Inventory Status
# - System Health
 
# Create custom dashboard
$ hyperfold dashboard create --name="Sales Overview" \
  --widgets='[
    {"type": "counter", "metric": "sessions_today"},
    {"type": "counter", "metric": "revenue_today"},
    {"type": "chart", "metric": "conversion_rate", "period": "7d"},
    {"type": "table", "metric": "top_products", "limit": 10}
  ]'
 
# Widget types:
# - counter: Single value with trend
# - chart: Time series visualization
# - table: Tabular data
# - gauge: Progress toward goal
# - heatmap: Activity patterns

Distributed Tracing

Trace requests across services:

bash
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
# View traces for a session
$ hyperfold traces get --session=sess_abc123
 
TRACE: sess_abc123
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
[0ms] Session started
  │   buyer: openai_gpt4_buyer
  │   customer: cust_xyz (Gold)
  │
[45ms] ├─ ACP Search "waterproof running shoes"
  │   │   products_found: 8
  │   │   semantic_confidence: 0.91
  │   │
[120ms] │   └─ Vector Search (Vertex AI)
  │       │   query_embedding: 1536 dims
  │       │   results: 8
  │       │   latency: 75ms
  │
[890ms] ├─ ACP Quote prod_aero_x2 @ $120
  │   │   context: {tier: gold, history: 12 orders}
  │   │
[920ms] │   ├─ Get Product (Firestore)
  │   │   │   cache_hit: true
  │   │   │   latency: 12ms
  │   │
[980ms] │   ├─ Calculate Dynamic Price
  │   │   │   floor: $92, target: $162
  │   │   │   adjustments: [gold -10%, competitor -0%]
  │   │
[1050ms] │   ├─ LLM Decision (OpenAI)
  │   │   │   tokens: 847
  │   │   │   decision: counter_offer
  │   │   │   latency: 130ms
  │   │
[1100ms] │   └─ Response: counter_offer @ $162
  │
[32000ms] ├─ ACP Checkout Finalize
  │   │
[32200ms] │   ├─ Process Payment (Stripe)
  │   │   │   spt_token: spt_live_xxx
  │   │   │   amount: $162.00
  │   │   │   status: succeeded
  │   │
[32500ms] │   └─ Create Order
  │       │   order_id: order_789
  │       │   status: confirmed
  │
[32550ms] └─ Session complete
            outcome: conversion
            revenue: $162.00
            duration: 32.5s

Alerting

Set up proactive alerting:

bash
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
# Configure alerts
$ hyperfold alerts create \
  --name="High Error Rate" \
  --condition="error_rate > 5%" \
  --window="5m" \
  --channels="slack:#alerts,email:ops@company.com"
 
> Alert created: alert_xyz123
 
# Alert conditions:
# - error_rate > X%
# - response_time_p95 > Xms
# - conversion_rate < X%
# - revenue_24h < $X
# - inventory_low (product specific)
 
# List active alerts
$ hyperfold alerts list
 
ACTIVE ALERTS
 
NAME                CONDITION              STATUS    LAST TRIGGERED
High Error Rate     error_rate > 5%        OK        Never
Slow Response       p95 > 500ms            OK        Dec 18, 14:30
Low Conversion      conversion < 20%       ALERT     Now
 
# View alert history
$ hyperfold alerts history --alert="High Error Rate"
 
ALERT HISTORY: High Error Rate
 
DATE                STATUS    DURATION   DETAILS
Dec 15, 09:15       ALERT     12 min     Peak: 8.2%, Root cause: DB timeout
Dec 10, 14:30       ALERT     3 min      Peak: 5.4%, Transient spike
Dec 05, 11:00       ALERT     45 min     Peak: 12%, Deployment issue
 
# Acknowledge alert
$ hyperfold alerts ack alert_xyz123 --message="Investigating"
View detailed logs with Log Analysis.