Build a sample AI stack and see where records may be created, retained, duplicated, or controlled by third parties.
For legal, risk, and insurance review, this helps answer the first practical question: where could the records be?
New to custody exposure? Start with a demo stack.
Pick a common AI workflow and Expose will show where prompts, outputs, logs, memory, telemetry, and admin records may be retained across vendors.
1. Choose workflow
Chat, RAG, voice, eval, or agent workflows create different records.
2. Choose deployment
Consumer defaults, enterprise terms, and ZDR APIs change custody assumptions.
3. Select vendors
Each vendor may hold a different piece of the record trail.
4. Read exposure
The result maps records, custodians, persistence risks, and follow-up questions.
Workflow
Deployment Configuration
Consumer
End-user interaction with platform defaults. History on, training opt-in varies, no enterprise terms.
Reduced
Feedback
Enterprise
API/ZDR
Agentic
Multi-Vendor
Select Your Vendors
⬢AI Provider
OpenAIASSESSED
AnthropicASSESSED
Google AIASSESSED
AWS BedrockASSESSED
◎Observability
DatadogASSESSED
SentryASSESSED
LLM TracingASSESSED
HeliconeASSESSED
OpenTelemetryASSESSED
◆Vector DB
PineconeASSESSED
ChromaDBASSESSED
■Database
SupabaseASSESSED
MongoDBASSESSED
◈Analytics
SegmentASSESSED
Session ReplayASSESSED
⬣Agent Framework
LangChainASSESSED
CrewAIASSESSED
▬Communications
TwilioASSESSED
⚙Auth
ClerkASSESSED
◈Payments
StripeASSESSED
Deployment Context
Matter Scope
Narrow the current stack analysis to a live legal event and generate response-oriented priorities.
←
Start by loading a demo stack or selecting vendors manually.
Expose will turn the selected stack into a custody map: which systems may hold AI records, which records may persist, which vendors may be independent custodians, and what questions a legal, insurance, or compliance reviewer should ask first.