phyla vs Gemini Enterprise Agent Platform: config vs cloud
Gemini Enterprise Agent Platform (formerly Vertex AI) is Google's cloud for building AI agents. phyla configures the agents you code with. How they differ.
Google’s Gemini Enterprise Agent Platform — the 2026 rebrand and evolution of Vertex AI — and phyla both surface under “AI agent platform,” but they answer different questions. The Agent Platform is Google Cloud’s environment for building, hosting, and serving AI agents. phyla configures the AI coding agents you already use to write software. They occupy different layers of the stack, so the real decision is which layer you’re shopping for, not which product wins.
What the Gemini Enterprise Agent Platform does
The Gemini Enterprise Agent Platform (formerly Google Vertex AI) is Google Cloud’s unified machine learning and generative AI platform. Through Model Garden it offers access to Gemini, Anthropic’s Claude, Llama, and 200+ other models behind a managed API. On top of model serving, it provides the tooling to build production AI systems: Agent Builder and the Agent Development Kit (ADK) for orchestrating agents, an Agent Engine runtime for deployment and memory, RAG and grounding for retrieval, plus evaluation, tuning, and MLOps infrastructure.
You consume it through the Google Cloud SDK, APIs, and console. It’s infrastructure for building and operating your own AI applications and agents, with Google handling scaling, security, and compliance. The audience is teams shipping AI features inside their own products.
Naming note: Google rebranded Vertex AI to the Gemini Enterprise Agent Platform in 2026, and “Vertex AI” no longer appears as a standalone service in the Cloud Console. The capabilities described here are the same ones teams knew as Vertex AI, now delivered under the Agent Platform.
What phyla does
phyla doesn’t host models or run agents in the cloud. It’s a visual configuration builder for the AI coding agents developers use every day — Claude Code, Cursor, GitHub Copilot, and others. You design how your agent behaves — project knowledge, conventions, skills, reusable commands — and export that configuration to whichever tools your team uses.
phyla lives at the developer-experience layer. It isn’t a competitor to Google’s Agent Platform; the two solve unrelated problems.
How they compare
| Gemini Enterprise Agent Platform | phyla | |
|---|---|---|
| What it is | Cloud ML & AI agent platform | Agent configuration builder |
| Primary job | Host models, build & run AI agents | Configure your coding agents |
| Layer | Infrastructure / PaaS | Developer experience |
| Who uses it | ML & platform teams building AI features | Developers using AI coding tools |
| Interface | Google Cloud SDK, API, console | Visual builder, exports to config files |
| Relationship to coding agents | A model backend they can run on | The config layer on top of them |
| Cost | Google Cloud usage-based pricing | Free to start |
The real question: are you building an app or configuring your tools?
That distinction resolves the comparison.
If you’re building an AI product — an agent, an assistant, a RAG system — and need managed models, orchestration, and MLOps, that’s what the Agent Platform is for. phyla doesn’t touch it.
If you’re configuring the AI agents you code with — getting Claude Code to follow your conventions, giving Cursor the right context, aligning Copilot with your standards — that’s phyla’s job, and the Agent Platform doesn’t touch it.
They also compose: Claude or Gemini models served through the platform can power coding agents, while phyla manages how those agents behave. Where you host a model and how you configure your coding agent are separate choices.
The platform’s “configuration” is application configuration
The word “configure” means something different on each side. On the Agent Platform you configure applications and agents for production: agent tools and orchestration, data stores and grounding sources, deployment endpoints, model parameters, and IAM. These shape how a deployed AI system behaves at runtime.
phyla configures coding agents: project context, behavioral rules, skills, and reusable commands that shape how an AI assistant helps you write code. Different artifacts, different audience, different lifecycle.
When the Gemini Enterprise Agent Platform is the right tool
Reach for the Agent Platform when:
- You’re building AI features into your own product and need managed model serving
- You want Gemini, Claude, and other models behind one platform with Google-grade scaling
- You need agent orchestration, RAG, evaluation, and tuning as managed services
- You’re standardized on Google Cloud and need its compliance and data controls
When phyla adds value
phyla is the right tool — whether or not you also use Google’s Agent Platform — when:
- You use AI coding agents like Claude Code, Cursor, or Copilot and want them configured well
- You use more than one tool and want a single source of truth for agent behavior
- Your team needs consistent agent conventions across projects and tools
- You want a visual interface for designing configurations instead of hand-editing rule files
- You want community skills to build from rather than starting empty
Using both together
The two stack naturally because they sit at different layers:
- Build and host your production AI application on the Gemini Enterprise Agent Platform (models, agents, grounding)
- Use coding agents like Claude Code or Cursor to write that application
- Configure those coding agents in phyla so they understand your codebase and conventions
- Export the configs to every tool your team uses and commit them to the repo
The Agent Platform runs your product; phyla tunes the tools you build it with.
Get started
Try phyla free — no account needed. Build your agent configuration visually and export to Claude Code, Cursor, or GitHub Copilot. It works the same no matter which cloud — Google’s Agent Platform, Bedrock, or Azure — serves the models underneath.