Disposition governance for enterprise agents

Turn agent activity into accountable outcomes.

ActiveSense gives enterprise AI teams a signed, replayable decision record for every governed agent action: authorized, refused, escalated, or accepted.

Salesforce stays the system of record for the work. ActiveSense records the agent disposition.

The current proof is the Receipt path. A live Agentforce External Service Action producing a Receipt, and any AppExchange package, are next proof rather than completed proof.

See the Receipt Request Salesforce demo
Decision record Signed Receipts for authorization, refusal, escalation, and acceptance.
Replayable path Show what was decided, why, and what evidence was available.
Closure objective Receipts are evidence. Closure is the objective.

Why now?

AI created labor. Enterprises still cannot see it.

Usage

We can measure tokens, traces, and spend.

Activity

We can see actions, commits, and workflow executions.

Outcome Gap

We still struggle to connect agent activity to accepted business outcomes.

The button worked. The decision did not.

The missing record

Logs record activity. Receipts record decisions.

Logs

  • Tool called
  • Request sent
  • HTTP 200
  • Workflow completed

Disposition Receipt

  • Authorized
  • Refused
  • Escalated
  • Executed
  • Accepted

Why decisions need proof

Action occurred. Closure unknown.

Enterprise AI does not fail only when systems break. It fails when actions occur and nobody can determine whether the loop eventually closed.

Refund

Refund approved, but never issued.

Escalation

Escalation created, but never accepted.

Deployment

Deployment approved, but never verified.

Export

Export refused, but never followed up.

Receipts make closure investigable.

Salesforce fit

Built beside Salesforce, not over it.

Agentforce defines and executes customer work. ActiveSense records the governed disposition so agent activity becomes accountable.

Salesforce / Agentforce

  • Action primitive
  • Customer record
  • Workflow execution
  • Semantic model
  • Platform audit

ActiveSense

  • Disposition primitive
  • Signed Receipt
  • Refusal, authorization, and escalation record
  • Replayable decision path
  • Accepted-outcome provenance

How it works

Every action ends with a disposition.

The path is intentionally small: an agent proposes or triggers action, ActiveSense checks authority and context, and the result becomes a signed, replayable Receipt.

01Proposed action
02Authority check
03Authorized
04Refused
05Escalated
06Accepted

Demo walkthrough

A governed action becomes a Receipt.

A Salesforce callout sends org and actor context to ActiveSense. If required Case context is missing, ActiveSense refuses, asks one concrete question, and records the disposition.

Agent proposes or triggers an action

Salesforce sends org identity, actor identity, source, and governed scope.

ActiveSense checks authority and context

Missing context becomes one concrete question instead of an unsafe route.

Disposition is recorded

The response includes receipt id, tenant lineage, disposition, and signed status.

Receipt can be verified and replayed

Replay shows the missing Case pointer and the exact question ActiveSense asked.

Receipt lookup
receipt_id: 3a9bf4...5aa3b0
tenant_id: 00Dfj...2cEAD
salesforce_org_id: 00Dfj...2cEAD
source: salesforce
disposition: authority_absent
issued_at: 2026-05-31T00:20:29Z
signed: true
Verify
signature_present: true
signature_valid: true
key_id: ce-receipts-prod-1
pressure_kind: missing_case_pointer
Replay endpoint
GET /control/receipts/{id}/replay
events: signal, pressure, question, receipt, verify
chain_length: 1
parent_status: root
chain_all_valid: true
question: Which Salesforce Case should this signal attach to?
signature_valid: true
Tenant isolation
correct tenant token: 200
spoofed tenant header: 403
foreign tenant token: 404

Enterprise examples

The same primitive across governed work.

Refund request

Agent proposes refund.

↓

ActiveSense checks authority.

↓

Receipt records whether the refund is authorized.

Customer export

Agent requests export.

↓

ActiveSense checks data sensitivity.

↓

Receipt records refusal when export is not allowed.

Escalation case

Agent proposes action outside authority.

↓

ActiveSense identifies the boundary.

↓

Receipt records escalation for human review.

Trust & verification

Receipts, not logs.

Logs say an event happened. ActiveSense Receipts say what was decided, why it was allowed or refused, and whether that decision can be verified.

  • Signed decision record Receipts include tenant scope, source context, disposition, and signing metadata.
  • Replayable context Operators can reconstruct the evidence and question that led to a refusal or escalation.
  • Accepted-outcome provenance Preserve the evidence trail connecting governed decisions to eventual outcomes.

Accepted outcome

Completion is not acceptance.

An agent can complete a task without producing a trusted business outcome. ActiveSense records whether work was merely completed or actually accepted.

Salesforce ecosystem

Salesforce owns the work. ActiveSense records the disposition.

Salesforce

Agentforce · Data Cloud · Trust Layer · CRM Records

↓

ActiveSense

Disposition Receipts

↓

Enterprise Audit / Governance / Accountability

The horizon

From accountable actions to coordinated systems.

Actions are owned by applications. Decisions span them. A refund touches Salesforce; its approval touched Slack; its fulfillment touched another system. As ActiveSense records the disposition of each governed action, those Receipts become a single, portable record across your systems — not just inside one. That cross-system accountability is the foundation of the coordination layer for the agentic enterprise: one place that knows what was decided, governed it, and can prove it, everywhere your agents act.

Tenant-scoped control API
GET /control/tenant
GET /control/boards
GET /control/receipts/{id}
GET /control/receipts/{id}/verify

tenant derived from bearer token
correct tenant read: 200
header spoof rejected: 403

Tenant isolation

Customer scope is enforced before data is returned.

ActiveSense maps authenticated access to tenant scope and rejects mismatches. A foreign tenant cannot read another tenant's Receipt payload through the Receipt API.

Economics

An additive layer beside Agentforce usage.

ActiveSense does not reduce Flex Credit usage or replace Salesforce controls. It helps enterprises trust, expand, and account for governed agent work.

Outcome evidence

Show which governed actions were authorized, refused, escalated, executed, or accepted — without claiming to calculate business value.

Expansion confidence

Give operators and buyers a decision record before more agents touch real customer work.

Evidence

Real Receipt examples, redacted.

Receipt lookup, verification, replay, and refusal evidence are available before a story has to be reconstructed from logs.

Real Receipt example
receipt_id: 3a9bf4...5aa3b0
source: salesforce
tenant_id: 00Dfj...2cEAD
disposition: authority_absent
signed: true
Verification example
GET /control/receipts/{id}/verify
signature_present: true
signature_valid: true
key_id: ce-receipts-prod-1
Replay example
GET /control/receipts/{id}/replay
events: signal, pressure, question, receipt
question: Which Salesforce Case should this signal attach to?
chain_all_valid: true
Refusal example
disposition: authority_absent
pressure_kind: missing_case_pointer
action_result: no unsafe route
next_step: ask one concrete question

FAQ

What enterprise buyers usually ask first.

Is ActiveSense a Salesforce replacement?

No. Salesforce remains the system of record for customer work. ActiveSense records the agent disposition around governed work.

How is this different from logs?

Logs record events. ActiveSense produces signed Receipts for governed decisions and refusals, with tenant, source, disposition, evidence pointers, and replay context.

How is this different from Salesforce Trust Layer?

Salesforce Trust Layer protects model interaction inside Salesforce. ActiveSense adds an independent disposition record after an agent proposes or performs governed work.

Has Agentforce already invoked ActiveSense through an External Service Action?

Not yet. The Receipt path is proven; one live Agentforce External Service Action producing a signed Receipt is the next proof.

Is ActiveSense on AppExchange?

Not yet. ActiveSense is being framed beside Salesforce, but an AppExchange listing or managed package should not be assumed from the current proof surface.

What does ActiveSense store?

ActiveSense stores Receipts, lineage pointers, tenant scope, disposition metadata, signing metadata, and replay evidence. It is not a CRM clone or data warehouse.

Can customers verify receipts independently?

Yes. Receipts include signing metadata and can be verified through the Receipt API or operator tooling.

Why are Receipts useful after a decision is made?

Because important outcomes often emerge days or months later. Receipts provide the evidence trail needed to reconstruct how a governed decision contributed to eventual closure.

What happens when an agent is uncertain?

Uncertainty becomes one concrete question attached to evidence. If required routing context is missing, ActiveSense records a signed refusal or escalation Receipt.

Make every governed agent action answerable.

See how ActiveSense records authorized, refused, escalated, and accepted dispositions beside Salesforce and Agentforce.

Not connected to our CRM yet — email hello@activesense.ai.

ActiveSense is a disposition governance system for enterprise AI.