| Intake: events from dispensing and service systemsAgree in discovery what your PMR and service IT suppliers can send as events, and what has to be read on a schedule. | API Gateway, EventBridge | API Management, Event Grid | Kong or your existing API gateway; Apache Kafka |
|---|
| Text messagesAmazon Pinpoint reached end of support on 30 October 2026; new builds use AWS End User Messaging. Use a sender that can receive replies, because a one-way alphanumeric sender can't take a 'STOP' text. | AWS End User Messaging SMS | Azure Communication Services SMS (UK data location) | A UK SMS gateway through your API layer |
|---|
| Automated booking callsNo predictive dialling: Ofcom's persistent misuse policy covers silent and abandoned calls. Test speech recognition on UK accents and older callers. | Amazon Connect (London) with Amazon Lex and Amazon Polly | Azure Communication Services Call Automation, Azure Speech in Foundry Tools | Open-weight speech models (e.g. Whisper) with an open-source telephony platform (e.g. Asterisk) |
|---|
| NHS App and NHS NotifyNHS Notify sends to the NHS App first, then text, then letter. Work to bring it to community pharmacy systems is under way but not generally available, so design the channel as a slot to fill, not a dependency. | Through your service IT supplier, once offered | Through your service IT supplier, once offered | Through your service IT supplier, once offered |
|---|
| Contact modelPredicts the best time and channel and who may need a call from a person. It orders and routes invitations; it never removes an eligible patient from the list. | SageMaker AI | Azure Machine Learning | MLflow and Kubeflow on Kubernetes |
|---|
| Template filling and reply sortingChooses an approved template and fills its slots, and sorts replies into booking, rescheduling, opt-out or clinical. Both clouds offer a choice of models, including Claude. | Amazon Bedrock | Microsoft Foundry | Open-weight models served with vLLM on your own GPUs, or IONOS AI Model Hub |
|---|
| Agent runtimeThe original Bedrock Agents closed to new customers on 30 July 2026; new builds use AgentCore. | Amazon Bedrock AgentCore | Foundry Agent Service | An open-source agent framework (e.g. LangGraph) on Kubernetes |
|---|
| WorkflowHolds each patient's journey: window opens, invite, reminder, call, booked, or handed to the branch. | AWS Step Functions | Logic Apps or Durable Functions | Temporal, or n8n |
|---|
| GuardrailsA final check before sending: the message must match an approved template exactly apart from its slots, with no medicine names, conditions or promotions. | Bedrock Guardrails | Foundry guardrails (Azure AI Content Safety) | Open-source guardrails (e.g. NeMo Guardrails, Llama Guard) |
|---|
| Platform | Managed AWS services | Managed Azure services | Kubernetes on-prem (e.g. OpenShift, Rancher), or IONOS Managed Kubernetes |
|---|
| Identity, keys, data protectionMobile numbers are personal data in their own right. Keep them in the register, not in prompts or logs. | IAM Identity Center, KMS, Macie | Entra ID, Key Vault, Purview | Keycloak or your existing Active Directory; OpenBao or HashiCorp Vault; an HSM |
|---|
| Monitoring and auditLog every message and call: template and version, why the patient was chosen, delivery receipt, reply and who handled it. | CloudWatch, CloudTrail | Azure Monitor, Log Analytics | Prometheus, Grafana, OpenTelemetry |
|---|
| UK data residencyAzure Communication Services lets you choose a UK data location; check where calls and messages are processed for your setup. | eu-west-2 (London) | UK South / UK West | Your own UK data centre, or a UK or EU provider (IONOS AI Model Hub runs in Germany) |
|---|