The immune system for trusted AI
Detect drift, score reflex loops, and ship recovery you can trace in GhostLog. Co-Pilot narrates on demand—grounded in system state, not vibes.
Your AI now has a measurable reputation. Higher trust unlocks stronger guarantees, better marketplace placement, and lower execution friction.
Trust boundaries you can operate, not just monitor
The product loop is detection → interpretation → action. Reflex and GhostLog hold the facts; recovery services propose moves; Co-Pilot explains and steers without replacing that stack.
Built structured-first so AI spend scales linearly
Narration is optional and capped per org (UTC day); deterministic paths stay hot. Engineering gates live in apps/dashboard/docs/cost-readiness-sprint-checklist.md (sprint companion + Definition of Done for the cost layer).
Visibility, narration, then guided action
Start on structured intelligence (near-zero marginal narration cost), add Pro when you need enforcement plus higher Co-Pilot limits, move to Trust+ when TrustGraph depth and action wiring matter.
Each plan includes monthly pools for agent runs, trust evaluations, and GhostLog steps, plus daily Co-Pilot narration caps (UTC). Heavier systems draw down pools first; incremental usage can be metered beyond that. Nothing is hidden — you are aligned to what your agents actually execute.
Rough monthly exposure vs. plan cost
Model incident cost and engineering time spent on trust fires. This is a directional estimate, not financial advice.
Starter
Reflex scoring, GhostLog, and recovery-aware signals with deterministic explanations first. Co-Pilot uses LLM only when you ask.
Start freeCore limits apply · Stripe checkout when enabled
Next unlock: Upgrade to Pro for enforcement, deeper execution visibility, and higher narration budget.
- ✓Reflex scoring + trust signals
- ✓GhostLog + structured event history
- ✓Recovery guidance (deterministic stack)
- ✓Co-Pilot: 20 LLM narrations / org / day (UTC); structured insights unlimited
- ✓Org-scoped dashboard
Pro
Production trust ops: enforcement paths, replay/drift workflows, and LLM narration for summaries and next-step copy—still grounded in your traces.
Billed monthly or annually · Metered overage on select units
Next unlock: Trust+ adds TrustGraph compute pools, higher narration caps, and guided action wiring.
- ✓Everything in Starter
- ✓Trust Lock enforcement
- ✓Replay + drift detection
- ✓Trust recovery workflows + API access
- ✓Co-Pilot: 200 LLM narrations / org / day (UTC); structured insights unlimited
Trust+
For teams proving reliability externally: guarantees, forecasting, TrustGraph-scale compute, and command-layer suggestions mapped to real APIs.
Highest leverage tier for partner-facing proof
Next unlock: Enterprise adds compliance mapping, attestation programs, and multi-org trust contracts.
- ✓Everything in Pro
- ✓Trust guarantees + forecasting
- ✓TrustGraph compute (metered included pool)
- ✓Marketplace-ready trust posture
- ✓Co-Pilot: 1000 LLM narrations / org / day (UTC); structured insights unlimited
Enterprise
Regulated and platform rollouts: dedicated trust architecture, audit-friendly exports, custom workflows, and SLAs.
Talk to usScoped pilots typical in 1–2 weeks
- ✓Everything in Trust+
- ✓Co-Pilot: unlimited LLM narration (enterprise posture)
- ✓Compliance mapping + attestation
- ✓Advanced routing and policy programs
- ✓Dedicated onboarding + support
Compare trust capabilities
Move from visibility → enforcement → guarantees → monetized trust.
| Feature | Starter | Pro | Trust+ | Enterprise |
|---|---|---|---|---|
| Trust Score | ✓ | ✓ | ✓ | ✓ |
| Co-Pilot structured insights (default) | ✓ | ✓ | ✓ | ✓ |
| TrustGraph compute (included pool) | — | — | ✓ | ✓ |
| Trust Lock enforcement | — | ✓ | ✓ | ✓ |
| Replay + drift detection | — | ✓ | ✓ | ✓ |
| Trust recovery | — | ✓ | ✓ | ✓ |
| Predictive trust (forecasting) | — | — | ✓ | ✓ |
| Trust guarantees | — | — | ✓ | ✓ |
| Compliance mapping | — | — | — | ✓ |
| Public attestation | — | — | ✓ | ✓ |
| Marketplace eligibility | — | — | ✓ | ✓ |
| Dynamic pricing + routing | — | — | — | ✓ |
| Dedicated support | — | — | — | ✓ |
What you actually get
Artifacts come from execution data: scores, recovery outcomes, and GhostLog lineage—not hand-wavy “AI analysis.”
Add-ons that stay margin-safe
Optional lifts for narration spikes, guarded automation, deeper GhostLog analytics, and org-specific Co-Pilot phrasing—priced as packages once billing hooks land.
Raise daily Co-Pilot LLM caps for spike weeks or executive reviews without changing core architecture.
Toggle guarded execution for approved recovery paths: rate limits, cooldowns, and GhostLog attribution required.
Deeper GhostLog aggregates, drift baselines, and scheduled risk digests—mostly structured compute, optional narration pass.
Org-specific narration templates and terminology so Co-Pilot matches your policies and customer language.
- ✕Silent AI failures
- ✕Undetected hallucinations and drift
- ✕No audit trail for high-stakes actions
- ✕No proof for customers, partners, or regulators
- ✓Every action is tracked, scored, and validated
- ✓Trust is enforced across execution paths
- ✓Failures trigger recovery workflows automatically
- ✓Exportable guarantees make reliability provable
Operate trust like infrastructure
Buyers want evidence: what broke, what recovered, and what policy blocked. Recon standardizes that narrative so GTM, security, and eng share one system of record.
Questions teams ask before adopting
How do you keep AI / LLM costs predictable?
Recon defaults to structured intelligence (Reflex, GhostLog, recovery summaries). Co-Pilot uses an LLM as a narration layer on top of that data, with per-org daily caps and ledger-backed metering—so you are not billed token-first.
How is this different from AI observability tools?
Observability tells you what happened. Recon enforces behavior, scores trust, runs recovery workflows, and produces artifacts you can show to customers and auditors—not just charts.
Do I need this if my AI already works?
Most systems appear to work until they fail silently. Recon makes trust measurable and prevents hidden failures from compounding.
How fast can we get value?
You will see a Trust Score and key signals within minutes. Enforcement and recovery layers activate as you integrate.
When should we upgrade to Trust+?
When trust becomes customer-facing. If you need to prove reliability or differentiate in the market, Trust+ is the inflection point.
Is this only for enterprise teams?
No. Starter and Pro are designed for developers and growing teams. Enterprise adds compliance, attestation, and marketplace leverage at scale.
Does trust actually impact performance or cost?
Yes. Higher trust unlocks better routing, lower friction, improved pricing, and stronger marketplace positioning.
Ship AI with a reflex loop you can defend
Start on structured intelligence, add narration where it helps, and scale action only with guardrails. Same stack from Starter to Enterprise.