We’re partnering with CrewAI to give new developers a friction-free path to building faster, more efficient AI crews.
Agentic AI is quickly becoming the mainstream way to deploy AI into production. However, for near real-time use cases, agents that require chaining many AI inference requests become impractical due to the token speed of each and every request. The speed requests need to be much faster in order to achieve response times in seconds, not minutes. At SambaNova, we are dedicated to making models run fast and efficient with our RDUs to drive upcoming agentic use cases.
That’s why we’re incredibly excited to announce our new partnership with CrewAI, the leading enterprise agentic AI platform for building collaborative AI agents. Together, we’re giving CrewAI AMP (agent management platform) users a free 7-day trial of models served on SambaCloud.
No initial setup. Just instant access to fast, cost-effective inference so you can focus on what you do best — building productive AI agents.
Why CrewAI AMP? Because the Future of Agentic AI Is Collaborative
A single AI agent is powerful, but a crew of agents working together is transformative. CrewAI enables exactly that — sophisticated multi-agent systems where AI “crew members” with specialized roles collaborate to solve complex tasks.
CrewAI AMP makes it incredibly easy for enterprises to develop their own agent integrated with their data and applications. With their no-code interface, you can describe what you would like your agent to do, which will automatically create a visual workflow that represents your agent. Below is an example of a crew with two agents: one that searches for news and another one that sends a daily digest email (a simplified version of ChatGPT Pulse).
With CrewAI, an agent is represented by an LLM, a goal, and a backstory. Agents are then assigned tasks, which are defined by a task description and an expected output. These tasks are assigned to agents and can either be stringed together sequentially to be completed like in the example above or assigned by a manager agent.
The visual editor makes it incredibly easy for enterprises to configure various agents without needing to code and integrate with a variety of different data sources. The most popular tools, such as Salesforce, Google Workspace and Slack, offer drag-and-drop components that anyone can use.
The SambaNova Difference: Built for the Workloads of Tomorrow
Available today on CrewAI AMP, enterprises can utilize any of the open-source models available on SambaCloud, including OpenAI’s gpt-oss and DeepSeek’s V3-0324 model. SambaCloud enables these models to run faster than any other platform and work extremely efficiently for agents.
One of the biggest requirements for agents to work successfully is accurate tool calling. Tool calling is the capability for a model to call a function with specified input parameters. In the example crew above, one agent is given access to search the web, while another agent is given access to send an email.
While most models today support function calling, the level of accuracy varies from model to model. Moreover, the implementation can vary from inference provider to inference provider. SambaNova tests our implementation rigorously for function calling to ensure the highest level of service with our optimized chat templates. We will be sharing more about our methodology in the coming weeks. Based on our testing, we are seeing significant improvements over accuracy compared to just the base model implementation on benchmarks like Berkeley Function Calling.
Function Calling Single Turn Simple Accuracy
| SambaNova | Base Model (from the leaderboard) | |
| DeepSeek-V3-0324 | 97.50% | 76.58% |
| Llama-4-Maverick-17B-128E-Instruct | 94.25% | 76.08% |
Join Us in Building the Future of Agents
Ready to get started? Sign up for CrewAI AMP to start your free trial!
Want to learn more about the technology powering your agents? Explore SambaCloud and see how we can help you scale.
SambaCloud is a powerful platform that enables developers to easily integrate the best open-source models with the fastest inference speeds. Powered by our state-of-the-art AI chip, the SN40L, SambaCloud provides a seamless and efficient way to build AI applications with fast inference. Get started today and experience the benefits of fast inference speeds, maximum accuracy, and an enhanced developer experience, in just three easy steps!