Skip to content

SYSTEM Cited by 1 source

GCP Vertex AI

Google Cloud Platform Vertex AI is GCP's unified managed ML platform that includes a hosted LLM model garden — Google's own Gemini family plus third-party models including Anthropic and others — with managed serving, fine-tuning, evaluation, and observability surfaces. Sibling system to Amazon Bedrock (AWS) and Amazon SageMaker (AWS).

Stub page on the wiki — expand as Vertex AI internals are disclosed in future ingests.

Wiki canonical role: enterprise multi-cloud LLM endpoint

The 2026-05-28 Slack AI multi-cloud retrospective is the wiki's first canonical disclosure of GCP Vertex AI as an enterprise multi-cloud LLM endpoint — specifically as the second cloud in Slack AI's Intelligent Routing Layer alongside AWS Bedrock.

Slack named four reasons for adding Vertex AI in early 2026 to their AWS-only stack (Source: sources/2026-05-28-slack-slack-ai-the-path-to-multi-cloud):

  1. Infrastructural redundancy & high availability — verbatim: "a multi-cloud footprint eliminates provider-level large scale infrastructural disruptions as a single point of failure. If an entire cloud ecosystem experiences a regional or platform-wide disruption, our traffic can be rerouted to a separate, healthy stack without service interruption."
  2. Model-to-feature optimisation — verbatim: "By expanding our catalog to include multiple models, we gained the ability to match the specific latent strengths of a model to the specific requirements of a feature. This granular optimization led to immediate performance gains: ~10% improvement in quality metrics for complex reasoning tasks. ~67% reduction in latency for high-velocity, low-token workloads."
  3. Access to innovation — verbatim: "The AI landscape moves at extreme velocity with frequent vendor exclusivity. Multi-cloud ensures we are ready to integrate with the latest breakthroughs regardless of where they are hosted while upholding our compliance, privacy, and security promises."
  4. Dynamic workload orchestration — verbatim: "Beyond simple failover, multiple providers allow for sophisticated traffic shaping. We can route requests based on real-time telemetry – evaluating not just provider health, but which endpoint offers the optimal performance profile for a given workload at that exact moment."

Integration friction (disclosed)

The Slack post discloses two named integration challenges that were resolved as cold-start engineering work for the Vertex AI addition:

  • Secretless authentication — explicit statement that Slack "solved cold start engineering hurdles by implementing secretless authentication" for cross-cloud access. Specific federation shape (workload identity federation? OIDC? short-lived service-account exchange?) not disclosed.
  • API normalisation layer — "translates disparate provider signals into a unified language for our application logic" — see patterns/api-normalization-layer-cross-provider.

Operational properties (disclosed)

  • Compliance + privacy + security promises maintained alongside AWS — Slack's framing is that the multi-cloud expansion was gated on cross-cloud parity for the existing enterprise compliance posture (FedRAMP Moderate at minimum, per Slack AI's earlier-phase requirements).
  • Vendor-exclusive models — frame for "access to innovation" implies state-of-the-art models that are exclusive to GCP at any given moment (e.g. some Gemini generation behaviours, certain Anthropic capabilities, etc.) — specific model SKUs not enumerated by Slack.
  • Cross-cloud routing — Slack's routing layer routes between Bedrock and Vertex AI based on metric-driven model selection per feature and on real-time health signals.

Open questions (from the Slack disclosure)

  • GCP region selection — which Vertex AI regions Slack uses, how regional data boundaries are honoured.
  • Per-cloud traffic share at Phase 4 — not disclosed.
  • Model SKUs Slack uses on Vertex AI — not enumerated.
  • Vertex AI capacity primitives — does Slack use Vertex AI Provisioned Throughput equivalent (Vertex AI offers "provisioned throughput" SKUs) or on-demand?
  • Vertex AI eval / monitoring tooling — does Slack use Vertex AI's own evaluation services or its in-house judging?

Seen in

  • sources/2026-05-28-slack-slack-ai-the-path-to-multi-cloud — the wiki's first canonical disclosure of GCP Vertex AI as an enterprise multi-cloud LLM endpoint alongside AWS Bedrock; named as the second cloud in Slack AI's Intelligent Routing Layer; four-reason rationale for multi-cloud expansion; secretless auth + API normalisation as cold-start integration work.
Last updated · 542 distilled / 1,571 read