DigitalOcean GradientAI Platform is now Generally Available

Over the past few months, thousands of developers explored the DigitalOcean GradientAI Platform, formerly known as the GenAI Platform, during its public preview. Their real-world usage and feedback directly shaped the GA experience, from building AI agents to integrating LLMs into production apps. Today, we're excited to announce that the GradientAI Platform is officially Generally Available (GA).

Today's release brings major improvements that make it easier to build, debug, and scale AI applications. New features like external data integration, agent traceability, customer conversation logs, and agent evaluation give you deeper visibility into how your agents behave, more control over their data and logic, and better ways to collaborate across teams-all without adding infrastructure overhead.

Since the initial release in January 2025, we've launched a wide range of enhancements driven by real-world adoption and evolving use cases. From foundational infrastructure upgrades to powerful new developer tools, here's what's changed between public preview and General Availability:

• External data integration: Connect external files and documents from Amazon S3 through API or control panel. This makes it easier to bring real-world context into your agents using RAG; for example, getting questions about internal documentation answered, generating summaries, or powering customer support agents.
• Agent traceability: New debugging tools show you how each request flows through your agent, including routing decisions, token counts, and system messages. This helps you pinpoint issues, improve accuracy, and optimize performance in production.
• Customer conversation logs: Capture and store agent interactions across your app, playground, or widget. With role-based access controls (RBAC) and data retention controls, you can export and analyze logs for analyzing user behavior, training better agents, or supporting audit and compliance needs.
• Agent evaluations: Run structured tests to assess your agents before deployment. You can use customizable templates to benchmark quality, safety, and accuracy. Agent evaluations are ideal for regression testing, tuning performance, or validating agents for high-stakes use cases.
• Workspaces: Workspaces act as containers for your agents and their resources, making it easy to isolate projects, manage team access, and collaborate across environments. Whether you're experimenting, supporting multiple clients, or working across teams, workspaces bring structure and clarity at scale.
• Improved agent management: We added versioning, usage insights, and integrated views for linked KBs, functions, and guardrails, all in a more intuitive agent management experience. This makes it easier to track changes, roll back versions, and manage complex agents confidently.
• More models, more flexibility: Support for newer models like GPT-4o, Claude 3.5, and DeepSeek R1 has been added. This lets you choose the best model for your specific workload, whether you're prioritizing speed, accuracy, or cost.

Not every AI use case requires a full agent. If you're just looking to integrate LLMs into your application, without the overhead of managing infrastructure or juggling API credentials, serverless inference on the GradientAI Platform gives you direct, flexible access to industry-leading models. It's ideal for developers who want to build quickly and stay in control.

Here's what you get with serverless inference on GradientAI Platform:

• Unified access to top LLMs like OpenAI, Anthropic, Mistral, and Meta
• A single API for all providers-no multiple keys or account setups
• Usage-based billing with no idle infrastructure or hidden fees
• Automatic scaling to handle unpredictable workloads
• Consistent endpoints and centralized monitoring across all model calls

Whether you're adding summarization to a SaaS tool, enhancing customer support, or building a content generation feature, serverless inference makes integration fast, clean, and scalable.

"The future of business is human-led and agent-operated. Humans articulate goals, and fleets of autonomous agents can execute on those goals," said Aaron Kettl, founder of Quickest. "With the GradientAI Platform, I can easily set up all of my agents and integrate them into my product with very little code, and I can upgrade and switch models anytime, so my product gets better every time a new model rolls out."

The GradientAI Platform is production-ready and built to grow with you, whether you're building agents to automate complex workflows or using serverless inference to add AI features to your app. This launch brings together everything you need to move fast, stay in control, and confidently ship AI to production.

And we're just getting started. In the months ahead, we'll be expanding the platform with smarter agent experiences, expanded evaluation tools, and new features that make building with AI even faster and more intuitive.

Get started on your agents ->

Call a model with serverless inference ->