Cloud integration considerations

AdaptCX is built on Google Cloud.

Google Cloud provides building blocks for AI-assisted customer experience, retrieval and document processing. AdaptCX integrations and controls must be evaluated for each implementation.

Request a walkthrough → See Sovereign Deployment
Built on
Gemini Enterprise for CX CX Agent Studio Agent Assist Vertex AI Google CCaaS Document AI Gemini GKE Autopilot CMEK
Latest from Google

What’s new in Google Cloud’s AI stack.

Google Cloud services evolve over time. Integration compatibility and the availability of new features require evaluation for the selected deployment.

Operational simplification

Why AdaptCX is the right way to run Google Governed AI architecture.

Google Cloud gives you raw building blocks. AdaptCX gives you the integrated platform — so you spend your team’s time on outcomes, not on plumbing.

Lifecycle

Versioned playbooks & model rotation

When Google ships a new Gemini variant, AdaptCX swaps it in behind a feature flag — with A/B comparison, rollback path, and audit-chain entries for every model swap.

Observability

One pane of glass for retrieval, generation, and audit

Cloud Monitoring dashboards out of the box. Latency stamps per stage. Cost attribution per tenant. Audit logs that map directly to interactions.

Compliance

CMEK end-to-end, audit-by-default

Key management and logging should be configured for each storage, retrieval and conversation component. Confirm the supported controls and record destinations.

Integration depth

Where AdaptCX and Google Cloud meet.

Vertex AI · RAG pipeline

Retrieval-augmented generation built on Vertex AI

AdaptCX’s retrieval pipeline runs on Vertex AI — text-embedding-004 for semantic embeddings, Gemini Flash for response generation to evaluate, and a tenant-isolated vector store for low-latency retrieval. Every component is permissions-aware, citation-grounded, and CMEK-encrypted.

  • text-embedding-004 with HNSW indexing for fast retrieval
  • Gemini Flash Lite for generative fallback design
  • Vertex AI Model Garden integration for evaluation
  • Source citation flowed through every response
RAG pipeline · per-query trace
utterance"what's my claim status"
embeddingtext-embedding-004 · 38ms
retrievalvector store · 47ms · 3 docs
playbookclaim_status_v3.1 (matched)
generationnot invoked · deterministic path
auditcited Policy 2024-114 §4.2
total312ms · CMEK-encrypted
Google CCaaS · voice integration

Native Google CCaaS integration for voice workloads

AdaptCX integrates directly with Google CCaaS — speech-to-text, conversation profile, Agent Assist, and CX Agent Studio. Every component is wired into the audit chain so spoken interactions get the same compliance treatment as text channels.

  • Google CCaaS speech-to-text with regulated-vocabulary tuning
  • Native Agent Assist UI module integration
  • Conversation Profile mapped to AdaptCX playbooks
  • Multi-language support across CCAI’s production languages
Google CCaaS voice flow
channelSIP · Telnyx → CCaaS
STTCCaaS · 187ms median
profileadaptcx_conversation_v2
routing→ AdaptCX playbook engine
surfaceAgent Assist UI
auditaudio + transcript + decision
storageCMEK in customer GCP project
Sovereign Deployment

AdaptCX inside your own Google Cloud organization

Sovereign Deployment is intended for customer-controlled cloud environments. Review service dependencies, permissions, key management, data flows and engineering access before implementation.

  • Tenant-isolated GKE Autopilot deployment with Shared VPC
  • CMEK across every storage and processing layer
  • Cloud Audit Logs integration for full trace
  • Workload Identity Federation for service auth
Sovereign Deployment · architecture
projectyour-org/adaptcx
networkShared VPC · your perimeter
computeGKE Autopilot · isolated namespace
datavector store · CMEK · your KMS
authWorkload Identity Federation
egresszero · stays in your perimeter
auditCloud Audit Logs in your project
Already on Google Cloud?

Run AdaptCX where you already are.

If your contact center is already on Google Cloud, adoption is faster — and Sovereign Deployment means we don’t move your data anywhere.

Request a demo → Sovereign Deployment details
Direct answers

Questions this page answers

Can generative AI run inside a customer's Google Cloud environment?

Yes. Organizations can design generative AI applications so application services, retrieval, data stores, identity controls, logging, and many supporting components run inside their own Google Cloud organization and projects.

The exact boundary depends on the Google Cloud services selected, network design, model endpoint, region, encryption approach, and governance requirements. The important design question is which controls the customer must own directly and which controls are provided by the cloud service.

Read the full answer →