Observability
Trace ledger built for AI
Every model call captured with cost, latency, risk flags, retrieval context, and user feedback.
- OTel-compatible SDK
- Saved views + alerts
- Long-form replay
NewTrust Score, public Trust Center, proof chain, and Annex IV files are live.Trust Center launch kit live.See what changed
Prove your AI is working as designed.
AegisRail connects production traces, eval gates, incidents, Trust Score, public Trust Center, proof chain, and audit-ready evidence into one operations console for AI SaaS teams.
Trust Score
0-100
Proof chain
SHA-256
Annex IV
Article 11
Frameworks supported
- SOC 2evidence ready
- ISO 27001control mapping
- EU AI ActArticle 9 / 15
- NIST AI RMFGOVERN / MAP
- GDPRDPA + DSR
New trust launch kit
Turn governance work into a sales asset.
Trust Score, public Trust Center, proof chain, and EU AI Act technical files turn the controls your team already runs into buyer-ready proof.
Public preview
Customer Trust Center
Trust Score
91
Grade A
Proof chain
verified
12 sealed events
Annex IV
ready
8 mapped sections
AI Trust Center
Public trust assetPublish a search-indexed page your customers can link to during security reviews, with Trust Score, systems, incidents, frameworks, and contact routing.
Tamper-evident proof chain
Audit integritySeal approvals, exports, eval runs, and incident resolutions into an append-only SHA-256 chain with a public verification endpoint.
Annex IV technical file
EU AI ActGenerate a versioned, article-mapped technical file from stored system data, ready before EU AI Act high-risk obligations apply; Annex III enforcement is set for December 2027 under the 2026 omnibus agreement.
Inside the platform, shipping today
The Aegis Flow
From runtime signal to audit proof, in one loop.
01
Observe
Capture traces, model responses, costs, latency, feedback, and runtime risk flags from production AI.
Drop the SDK in your serving layer. Traces arrive p95 sub-100ms with no sampling for high-risk paths.
14.2k
traces / day
Product
One platform. Four operating surfaces.
Each surface is independently useful and built to compose. Adopt one today; wire the rest in when you're ready.
Evals
Regression gates before launch
Run deterministic checks and Gemini-graded rubrics on every change. Block bad releases automatically.
- Scheduled suites
- Release gate API
- Rubric versioning
Governance
A control plane for AI trust
Map systems to owners, data classes, and review policies. Turn live posture into a shareable Trust Score.
- Trust Score grades
- Incident workflows
- Slack + webhook delivery
Evidence
Audit-ready proof, on demand
Package approvals, traces, policies, and technical files into tamper-evident proof for buyers and auditors.
- Proof chain manifest
- Public Trust Center
- One-click handoff
Ledger
Every trust event, sealed in order.
Approvals, exports, eval runs, and incident resolutions flow into a tamper-evident proof chain with a public verification manifest.
liveappend-only · checksum-verified
tr_2af91c7
flagAI Support Agent · negative feedback
risk flag · pii redaction · evaluator queued
ev_3c12a4b
passRefund Policy Evaluator · v3
42 / 44 cases passed · regression contained
bd_71e0fc9
sealEvidence Bundle · SOC 2 § CC7.2
sealed · checksum 9c4e2af3b1d
tc_8a91e30
livePublic Trust Center · Growth workspace
indexed · proof manifest linked · buyer-ready
tr_2af91d2
failOnboarding Agent · tool failure
3.1s · retry budget exhausted
in_55a8c12
openIncident IN-118 · drift on intent class
owner notified · timeline updated
ev_3c12a51
gatePII Redaction Rubric · v1
15 / 18 cases · 2 regressions flagged
ap_8d2f01a
okApproval · Control C-15 access review
reviewer: sec-lead · status: signed
tf_413799c
fileAnnex IV technical file · Support Copilot
Articles 9 / 10 / 12 / 14 / 15 / 72 mapped
tr_2af91c7
flagAI Support Agent · negative feedback
risk flag · pii redaction · evaluator queued
ev_3c12a4b
passRefund Policy Evaluator · v3
42 / 44 cases passed · regression contained
bd_71e0fc9
sealEvidence Bundle · SOC 2 § CC7.2
sealed · checksum 9c4e2af3b1d
tc_8a91e30
livePublic Trust Center · Growth workspace
indexed · proof manifest linked · buyer-ready
tr_2af91d2
failOnboarding Agent · tool failure
3.1s · retry budget exhausted
in_55a8c12
openIncident IN-118 · drift on intent class
owner notified · timeline updated
ev_3c12a51
gatePII Redaction Rubric · v1
15 / 18 cases · 2 regressions flagged
ap_8d2f01a
okApproval · Control C-15 access review
reviewer: sec-lead · status: signed
tf_413799c
fileAnnex IV technical file · Support Copilot
Articles 9 / 10 / 12 / 14 / 15 / 72 mapped
For your team
Built for everyone who owns AI in production.
Engineering
Ship AI without flying blind
- Trace every model call with latency, cost, retrieval context
- Block regressions at the release gate, not in prod
- Drop-in SDK; OTel-compatible
<100mstrace ingest p95 budget
Security
AI systems mapped, monitored, trusted
- Risk-tier every system; map owners and data classes
- Publish a customer-facing Trust Center when buyers ask
- Incident timelines with Slack + signed webhooks
A-Dtrust grade per system
Compliance
Audit and EU AI Act packets without the scramble
- Evidence bundles with proof chain checksums
- EU AI Act Annex IV technical files from stored data
- Export PDF + machine-readable in one click
Art. 910 · 12 · 14 · 15 · 72 mapped
Operating targets
The boring kind of impressive.
The budgets AegisRail is engineered to hit, end to end.
- ms · trace ingest p95 budget
- 0
- regressions past the gate
- 0
- point Trust Score
- 0
- click · evidence handoff
- 0
from SDK call to queryable trace
failing eval suites block the release API
A-D grade across every registered AI system
from auditor request to sealed bundle
Pricing
Start internal. Publish trust when it matters.
Every plan gets internal Trust Score and proof-chain integrity. Growth unlocks the public Trust Center and EU AI Act technical files. Scale adds signed evidence exports for enterprise reviews.
Founding pricing: launch rates are locked for 12 months for early workspaces.
Public trust features follow server-side entitlements, so downgrades remove public access automatically.
Plan
Free
Validate one AI workflow before inviting a team.
$0/month
- Internal Trust Score
- Tamper-evident proof chain
- Public AI Trust Center
- EU AI Act technical file
- Signed evidence exports
1 AI systems · 1 users · 5,000 traces/month
Plan
Starter
Small SaaS teams preparing for customer security reviews.
$29/month
- Internal Trust Score
- Tamper-evident proof chain
- Public AI Trust Center
- EU AI Act technical file
- Signed evidence exports
3 AI systems · 3 users · 25,000 traces/month
Recommended
Growth
Publish Trust Center and EU AI Act evidence across AI products.
$99/month
- Internal Trust Score
- Tamper-evident proof chain
- Public AI Trust Center
- EU AI Act technical file
- Signed evidence exports
10 AI systems · 10 users · 250,000 traces/month
Plan
Scale
Signed evidence exports and TrustOps scale for enterprise AI vendors.
$799/month
- Internal Trust Score
- Tamper-evident proof chain
- Public AI Trust Center
- EU AI Act technical file
- Signed evidence exports
Unlimited AI systems · Unlimited users · Unlimited traces/month
The thesis
Evidence shouldn't be a deliverable your team scrambles to assemble every quarter. It should be a byproduct of how you build.
observe · evaluate · govern · prove
Ready for AI governance
Replace scattered AI evidence with one customer-ready system.
Set up your workspace, drop the SDK in, and ship your first sealed evidence bundle in an afternoon.