Observe

See Everything. Store Nothing You Shouldn't.

Full metadata observability for every AI session — model, tokens, cost, latency, and anomalies. No prompts. No responses. No payload logging by default.

Real-time telemetry
30-day retention (Solo)
Metadata-only by default
Per-session drill-down

Every session, captured

Metadata fields sent per session. Never prompt or response text.

tsISO-8601 timestamp of session start / end
modelModel identifier (e.g. gpt-4o, claude-3-5-sonnet)
tokens_inPrompt tokens consumed this session
tokens_outCompletion tokens generated this session
latency_msWall-clock ms from first token to last
cost_usdCalculated cost using live pricing table
user_idHashed identity — never raw email in payload
project_idRepository / workspace identifier
session_idUnique UUID per session for drill-down
error_codeHTTP / model error code if session failed
Session detailsess_a4f92c
{
  "session_id": "sess_a4f92c1e",
  "ts":         "2026-05-29T14:32:01Z",
  "model":      "claude-3-5-sonnet-20241022",
  "tokens_in":  12401,
  "tokens_out": 3892,
  "latency_ms": 820,
  "cost_usd":   0.1843,
  "user_id":    "u_8f3e…  ← hashed",
  "project_id": "proj_forg-dashboard",
  "error_code": null
}
No prompt text. No response content. Metadata only.

Live session feed

Real-time ingestion — sessions appear within 50 ms of completion.

TimeUserModelTokensCostLatencyStatus
14:32:01alice@co.comclaude-3-5-sonnet12.4K$0.18820mscompleted
14:31:44bob@co.comgpt-4o48.1K$2.403.2scompleted
14:31:12carol@co.comclaude-haiku1.2K$0.02180mscompleted
14:30:58dan@co.comgpt-4o9.8K$0.491.1serror
14:30:33eve@co.comclaude-3-5-sonnet22.7K$0.341.9scompleted
14:29:47frank@co.comclaude-haiku3.5K$0.04310msbudget alert
14:29:11grace@co.comgpt-4o31.2K$1.562.6scompleted
14:28:55hank@co.comclaude-3-5-sonnet6.9K$0.10640mscompleted

Token usage by model this week

12.3M tokens across all models — last 7 days.

GPT-4o
42%
5.2M
Claude Sonnet
31%
3.8M
Claude Haiku
18%
2.2M
Others
9%
1.1M

Spot the unusual before it's expensive

FORG surfaces behavioural anomalies automatically — no rules to write.

CRITICAL

User spike 890% above 7-day average

user: eve@co.com — $340 spend in the last 2 hours. Session loop suspected.

WARNING

New model detected: claude-3-opus

Model not in team allowlist. 3 sessions fired before block rule applied.

INFO

Off-hours usage detected

47 sessions logged between 00:00–05:00 UTC. Review for unattended scripts.

Privacy-respecting visibility

Every role sees exactly what they should — no more, no less.

Individual developer
Own sessions only
Full session history for your own activity
Per-session cost and token breakdown
Model usage and latency drill-down
Export your own data as CSV or JSON
Team lead
Aggregate view — k-anonymity ≥ 5
Team-level token and cost aggregates
Model distribution across the team
Anomaly alerts across all members
No individual session drill-down
k-Anonymity rule
Aggregates require ≥ 5 team members. Groups smaller than 5 return HTTP 422 — protecting individual privacy at the API level.

Retention and export

Your data stays as long as you need it.

PlanRetentionExport formatsNotes
Solo30 daysCSV
Professional12 monthsCSV + JSON
Team24 monthsCSV / JSON / NDJSON
EnterpriseCustomAll formats + S3 pushBring your own bucket

Connect your observability stack

FORG speaks the protocols your infra team already trusts.

D
Datadog
available
G
Grafana
available
P
PagerDuty
coming soon
S
Slack
coming soon

Start observing your AI usage

Install the adapter in 60 seconds. Your first session appears in the dashboard within minutes — full metadata visibility, zero prompt logging.