◈ Intelligence

Intelligence Without Inference Lag

FORG's intelligence layer runs entirely on statistical ML — no LLM in the real-time path. Every anomaly, classification, and recommendation fires in microseconds, not seconds.

< 2ms
analysis latency
98.7%
anomaly detection rate
Dashboard-only
FORG Atlas
Zero LLMs
in real-time path
Get started at $19/month →Architecture docs

How FORG Intelligence works

Three layers, zero LLM latency. The Go agent collects signals and ships them encrypted to the Rule Engine. A statistical ML engine classifies each event — cost anomaly, model drift, off-hours usage — before any rule fires.

FORG Atlas lives entirely outside the real-time path. It runs as a batch process against your Supabase + pgvector index, surfacing natural-language answers in the dashboard only — never blocking a hook evaluation.

Statistical ML: isolation forests + z-score for anomaly detection
Cost classifier trained on 400K+ usage sessions
Model recommendation engine using task-type embeddings
FORG Atlas: pgvector + NL query, batch-only
No LLM sits in the critical evaluation path

┌─────────────────────────────────────────┐
│           FORG Agent (Go binary)        │
│   signal collector · zero intelligence  │
└──────────────────────┬──────────────────┘
                       │ encrypted signals
                       ▼
┌─────────────────────────────────────────┐
│       Statistical ML Engine             │
│  · cost classifier  · anomaly detector  │
│  · model usage profiler                 │
│  · drift monitor                        │
│  < 2ms per evaluation                   │
└──────────────────────┬──────────────────┘
                       │ labelled events
                       ▼
┌─────────────────────────────────────────┐
│         Rule Evaluation Layer           │
│  · profile lookup  · threshold check    │
│  · alert dispatch  · audit chain        │
└──────────────────────┬──────────────────┘
                       │ (dashboard only)
                       ▼
┌─────────────────────────────────────────┐
│           FORG Atlas (batch)              │
│  · pgvector index  · NL query engine    │
│  · grounded answers · usage analytics  │
│  NOT in real-time path                  │
└─────────────────────────────────────────┘
Dashboard-only · Batch processing · Not real-time

Natural language queries, after the fact

FORG Atlas answers questions about your historical usage data. It runs in the dashboard — never in the hook evaluation path.

QueryWhich team spent the most on GPT-4o last week?

Engineering led GPT-4o spend last week with $412.80 — 54% of org total. Breakdown by sub-team:

TeamSessionsTokens (M)Cost
Engineering1,2048.2$412.80
Product5433.1$155.00
ML Research3122.4$120.40
Design880.4$20.00
QueryShow me sessions that exceeded 50K tokens

3 sessions found in the last 30 days:

sess_8x2adan@acme.com84,200 tokgpt-4o$2.53May 21 09:14
sess_4k9csara@acme.com71,800 tokclaude-3-opus$5.39May 18 14:02
sess_2m1dci-bot@acme.com63,400 tokgpt-4o-mini$0.38May 16 02:31

Catch cost spikes before they escalate

Statistical anomaly detection fires in < 2ms per session evaluation.

criticalToday 09:14 UTC
Cost spike detected

user dan@acme.com consumed $84.20 in 23 minutes — 9.4× their 7-day average.

dan@acme.com+$79.40 vs baseline
warningYesterday 16:42 UTC
Unusual model switch

engineering team moved from Claude 3 Haiku → GPT-4o for code review tasks. Projected cost impact: +$320/mo.

engineering team+$320/mo projected
info2 days ago 02:31 UTC
Off-hours usage detected

3 automated sessions ran between 02:00–04:00 UTC with no rate limiting. No policy covers this window.

ci-bot@acme.com$12.80 unclassified

Right model for the task

FORG profiles each session by task type and surfaces the cost-optimal model with evidence.

Task type
Code generation
Recommended
Claude 3.5 Sonnet
Cost / 1K tok
$0.003
Est. savings/mo
$140

Highest pass@1 on HumanEval with 4× lower cost vs GPT-4o

Task type
Long-form writing
Recommended
Claude 3 Opus
Cost / 1K tok
$0.015
Est. savings/mo
$60

Superior coherence over 8K+ tokens; avoids repetition artifacts

🔍
Task type
Quick lookup / RAG
Recommended
Claude 3 Haiku
Cost / 1K tok
$0.00025
Est. savings/mo
$310

~1ms TTFT, 200× cheaper than GPT-4o for retrieval-only tasks

Without FORG vs with FORG

✗ Without FORG
Discover overspend in monthly invoice
Manual log triage takes hours
No model recommendations
Blind to model drift across teams
✓ With FORG
Real-time anomaly alert within seconds
ML-classified sessions, searchable instantly
Automated per-task model suggestions with ROI
Drift detection with projected cost impact
47ms
avg dashboard query
99.9%
uptime SLA
30 days
retention (Solo)
Unlimited
retention (Enterprise)

See your team's intelligence layer

Statistical ML anomaly detection, FORG Atlas dashboard queries, and model recommendations — all included.

Get started at $19/month →Read the docs