Observer

Free core. Paid modules.
Metered AI.

The plugin is open source and the base queue is free. Paid modules are billed monthly, per community. AI is metered by credits — you pay only for what gets classified.

The private beta is free. These are the prices when we ship 1.0.

The Core. Free forever.

Plugin, ingestion, manual queue, audit log. No card needed. No trial timer. Use it indefinitely for one community.

$0 · forever

Observer Core

Everything you need to start. Adds AI screening the moment you buy credits.

  • + Plugin (Apache 2.0)
  • + Discord centralized bot
  • + Event ingestion (100k/mo)
  • + Manual moderation queue
  • + Audit log (90d retention)
  • + Dashboard for 1 community
  • + Email support
  • + Public roadmap input
add AI: buy credits to enable classification · long-form · trust scoring

Paid modules.

Add only what you need. Monthly, per community. Cancel any time. Bundle below saves 30%.

Long-form pattern detection

$39 /mo · per community

+ AI credits

Catches grooming, coordinated raids, multi-account scams, and slow-burn harassment — patterns that span days, accounts, and channels.

  • +Detects across messages, not just within them
  • +Cross-account fingerprinting
  • +Tunable look-back window (default 5 days)

Trust & ActorRiskScore

$19 /mo · per community

+ AI credits

0–100 risk score per actor, recomputed hourly, with a transparent component breakdown. Threshold-based auto-actions.

  • +Component-weighted scoring (you tune)
  • +Per-actor history and trajectory
  • +Severity → action policy mapping

Transparency & compliance

$29 /mo · per community

DSA Article 24 reports, COPPA-age-flag tooling, UK OSA risk-assessment templates, per-user DSR (access + erasure) workflows.

  • +Automated DSA exports
  • +DSR portal for users
  • +Auditor read-only view

Appeals workflow

$19 /mo · per community

User-facing appeal form, senior-reviewer queue, audit-log forks per appeal, and feedback that retrains your classifier.

  • +User self-serve appeal portal
  • +Reviewer queue with SLA tracking
  • +Active learning from reversals

Custom rules engine

$19 /mo · per community

Deterministic rules alongside the AI classifier — for the cases where you know exactly what you want banned.

  • +Regex + structural matchers
  • +List-based catch terms
  • +Runs alongside AI screening

Studio

$99 /mo · per organization

For studios running multiple communities. Multi-server roll-up, team roles, SSO/SAML, audit federation, and priority support.

  • +Multiple communities under one org
  • +Team roles + SSO
  • +Federated audit log

Operator pack

All modules — $149 /mo

Long-form patterns, trust, transparency, appeals, custom rules, studio multi-community. Bundle saves ~30% vs adding them individually.

Start with the pack

AI credits.
You pay for what gets classified.

Every event your modules classify draws from a credit pool, so a quiet community pays little. A 22k-member Discord typically runs $20–$40 a month at Haiku-tier classification.

We pass model costs through at ~1.4× retail to cover prompt caching, inference orchestration, and fallback routing. The math is published; no markup surprises.

/billing/ai-credits model · use · rate
Haiku 4.5 Standard classification $0.50 / 1k events 80% of typical queue traffic. Prompt-cached system prompts.
Sonnet 4.6 Contextual classification $5 / 1k events Used for borderline cases and edge categories.
Sonnet 4.6 Long-form pattern run $0.10 / pattern check Each pattern evaluation across N messages.
Opus 4.7 Hard-edge cases $2 / 1k events Senior-reviewer escalations. Manual trigger only.

Credit packs · pre-pay for rebates

$25

~50k events

starter

$100

~250k events

small community

$500

~1.5M events

studio · 5% rebate

$2,000

~7M events

scale · 10% rebate

Common pricing questions.

Is there really a free forever tier? +
Yes. Plugin, ingestion, manual queue, audit log — free, indefinitely, no card. We make money on modules and AI credits. If you never enable AI or buy a module, you never pay us.
What's a "community"? +
One Discord server or one Minecraft network (multiple shards count as one) or one Roblox group. Modules and the per-community pricing apply per such unit. Studio orgs get federation across all of them.
What if I run out of credits? +
AI screening pauses. Events still ingest, still hit the manual queue, audit log keeps logging. You get an email + dashboard banner at 80%/95%/100% of pool. Auto-reload is opt-in and disabled by default.
Can I bring my own Anthropic API key? +
Yes — on the Studio tier. Your key, your bill, your rate limits. We meter usage in the dashboard but skip the credit reseller markup. Useful if you’ve already negotiated direct pricing with Anthropic.
Annual pricing? Discounts? +
Annual prepay: ~17% off (two months free). Non-profits and education: 50% off modules. Open-source projects: full modules sponsored, talk to us. Credit packs above $500 include rebates as listed.
Refunds? +
Monthly module charges: full refund within 14 days of first charge, pro-rated thereafter. Credit packs: unused credits refundable within 30 days. Nothing locked.

Free during the beta.

During the private beta you get full module access and a generous credit allowance, free, while we refine Observer with a small group of operators. Pricing starts at 1.0.

→ No credit card, no sales calls. A real person replies within the day.