BP technology workshop presenter path

Technology Workshop Demo Flow

Use one end-to-end duplicate-invoice event to answer the five questions technology leaders care about: is it real, how does it work, how does it integrate, how is it governed, and how does it fail safely.

SignalContextPolicyEvidenceDecisionHuman ApprovalExecutionReplay

Definition to use first

bp Sphere is a governed enterprise decision runtime that connects signals, context, policies, evidence, agents, humans, and systems of record into a single accountable execution fabric.

Primary demoP2P duplicate
Deep diveRuntime
ControlsPolicy + replay
Future bridgeContinuous close

1. Is it real?

Show the live P2P queue, case drawer, evidence pack, action surface, SAP context drawer, and replay. Avoid starting with diagrams.

2. How does it work?

After the business event lands, show agent activation, context retrieval, policy evaluation, evidence assembly, decision formation, and replay.

3. How does it integrate?

Show SAP ECC/S4/CFIN coexistence, Ariba, Databricks/UDP, OneData/Foundry, BlackLine, ServiceNow, and evidence storage as adapter contracts.

4. How is it governed?

Show Entra-backed agent identity, AI use-case approval, Material AI gate, kill switch, policy gates, human approval, audit trail, and revocation.

5. How does it fail safely?

Show fallback policy, human workflow, replay, recovery, and business continuity. AI is optional; continuity is not.

25 minute live demo sequence

One Business Event, Eight Proof Points

Every proof point links to an existing system page. This page is the presenter shell, not another architecture deck.

1

Signal Arrives

Open the live P2P queue and anchor on the duplicate invoice event. The point is that the demo starts from a business signal, not a chatbot prompt.

Open live P2P
2

Context Assembly

Show invoice, PO, receipt, prior payment, supplier history, and payment terms in the case drawer. This answers whether the agent is guessing.

Open adaptive analyst workspace
3

Policy Evaluation

Show duplicate-payment, authority, materiality, evidence, and payment-hold policy checks. This answers how AI is controlled.

Open P2P policy path
4

Evidence Pack

Open invoice, PO, receipt, prior payment, and supplier evidence through the Evidence Intelligence Runtime.

Open evidence
5

Recommendation

Show hold-payment recommendation, confidence, reasoning factors, and expected outcome before any action is taken.

Open case drawer
6

Human Approval

Show analyst/supervisor boundary: hold, escalate, request evidence, open supplier, open SAP context. Human accountability remains visible.

Open supervisor queue
7

Governed Execution

Show that write-back is controlled: draft action, SAP context, audit record, queue impact, no autonomous payment release.

Open SAP coexistence
8

Replay

Show Signal -> Context -> Policy -> Evidence -> Recommendation -> Human Approval -> SAP update as an inspectable replay path.

Open replay
Under the hood after the demo

Technology Deep Dive Order

TopicShowWhy it matters
Agent runtimeRuntime topologyState, coordination, observability, cost, ownership.
Data and integrationData strategyProtect SAP and use UDP/Databricks/OneData/Foundry correctly.
Identity and governanceEntra agent identityAgents are governed identities, not anonymous bots.
SAP coexistenceQuantum coexistenceECC, CFIN, and S/4 can coexist during rollout.
ResilienceFailure handlingFallback, recovery, human workflow, replay.