Key takeaways
- CoreWeave transformed its customer support within 90 days by embedding Cohere’s North directly into its Slack-based workflows.
- North’s automated triage and context-gathering improved routing accuracy, accelerated resolution times, and supported strong customer satisfaction scores.
- North empowered CoreWeave’s support staff and engineers to deliver faster, higher-quality support, paving the way for further agentic AI adoption.
The promise of AI agents to help companies improve their customer service is clear. So far, though, only a few organizations have turned that vision into reality.
CoreWeave, a global AI cloud leader based in the US, is acutely aware that solving customers’ problems fast is fundamental to success. The pioneers who rely on CoreWeave’s infrastructure and support include the world’s top AI labs and global enterprises. Their time is highly valuable; just an hour of downtime can be extremely expensive.
That’s why CoreWeave has built its support culture on the principle of "Direct to Expert," a commitment to connecting customers quickly with the engineers most qualified to solve their issues. It has a simple motto: Customer first, always.
“Every interaction has to protect their time and keep their systems running”
John Mancuso, VP of Customer Experience, CoreWeave
The initial challenge
Facing rapid growth in demand, CoreWeave wanted to streamline its support system by reducing the number of third-party tools, increasing automation, and reducing the amount of manual steps taken by customer support engineers (CSEs) to route issues to other engineering teams.
"There were too many barriers," says Ivy DeWitt, Senior Manager of Customer Support Engineering. "It was clunky and slow — the opposite of what we wanted."
To ensure that it stayed true to its motto, CoreWeave needed a more efficient workflow that would enhance its engineers' ability to deliver fast, expert service to customers at scale.
A 90-day transformation in customer support
CoreWeave turned to Cohere and North, our agentic AI platform. Working with Cohere’s solutions architects, CoreWeave revamped its customer support process in just 90 days, resulting in a more efficient and automated process that has led to faster routing and resolution times, stronger customer satisfaction, and compliance with CoreWeave’s privacy standards.
Three features of North stood out to the CoreWeave team and improved inefficiencies:
Integration with existing apps: Unlike many AI tools that are tied to a fixed chat UI, North can be integrated via API into existing business software, workflows, and automations. For CoreWeave, that meant they could embed North directly into their preferred support system: Slack.
Privacy and control: Because North would be privately deployed inside CoreWeave’s own data centers, sensitive customer data never leaves their security perimeter – a contrast to publicly hosted AI models, where data is less secure. This enables CoreWeave to address compliance requirements and retain control over its systems and capacity.
Model flexibility: North’s architecture supports multiple models and configurations, enabling CoreWeave to match the right model to each use case.
Over an initial two-day on-site, Cohere worked closely with CoreWeave’s engineering and business teams to map out the end-to-end process, identify the specialized knowledge required for the AI agents, and establish a timeline.
"Having buy-in from the right stakeholders from start to finish was the key part of a successful implementation," says Matthew Koscak, a Solutions Architect at Cohere.
Just three months later, CoreWeave had North deployed broadly in production.
“From the start, North’s ability to meet CoreWeave’s preference for a system that worked through APIs made it a great fit,” says Koscak.
CoreWeave’s new North-powered customer support
Previously, CoreWeave’s CSEs would manually piece together context for new tickets and follow up with the requester for missing information. They would then have to then submit requests through a separate support portal in FreshDesk, which would ultimately bring them back to Slack for internal troubleshooting through a separate automation.
With North, CSEs can trigger their internal support automation and receive relevant documentation and suggestions for resolution steps automatically — all without leaving Slack. These automations help save time, improve efficiency, and enable CSEs to focus on solving engineering issues.
Here’s how the North-powered customer support system works:
A new support issue arrives via Slack from a CoreWeave customer.
CSEs trigger the North triage bot in Slack.
North collects and pre-populates key customer context, such as region, cluster, and other customer-specific information from CoreWeave's backend record-keeping systems.
A CSE quickly reviews the ticketing information for accuracy, adds any missing nuance, and confirms the ticket.
A separate internal automation creates a Jira issue, and from there, the swarming channel is opened, and CoreWeave’s automations pull in any relevant Engineering teams based on North's recommendations.
The opening of the swarming channel triggers a separate North resolution agent, which pulls in CoreWeave's documentation and historical post-ticket-reviews to provide the best possible resolution.
The overall result: a more unified, streamlined workflow that enables CoreWeave engineers to hit the ground running with a comprehensive suggested resolution at hand.
By helping the support team triage and resolve issues, North enables CoreWeave to scale its engineering resources while solving problems faster and more accurately. Teams can stay within CoreWeave’s preferred support communication channel, Slack, and engineers can focus on what they do best: delivering rapid, expert solutions to customers.
The impact: Faster, more responsive, richer customer support
Within a few months of deployment, the CoreWeave team identified clear performance improvements.
Days saved for mean time to resolution: Customers’ problems are solved faster, thanks to better intake data, more detailed context, and fewer manual steps. On average, CoreWeave’s resolution time decreased from 4-8 days to 2-5 days due to improved internal procedures coupled with North automation.
4.9 - 5.0 customer satisfaction (CSAT) scores: These scores continued for the vast majority of customer support tickets after the first few months of deploying North.
Increase in routing accuracy: This measurement indicates that more customers are being connected to the right resource.
Beyond the raw numbers, the CoreWeave team has anecdotally noticed that support quality has improved as the engineers can now devote more time to solving problems rather than gathering information and switching between systems. The team says that the minor frustrations with support quality that customers used to voice have been absent since the new system launched. They’ve also been pleasantly surprised by North's strong stability record, since it needs to connect to multiple internal systems.
“For all of the systems we integrate with, it's been a very, very reliable platform,” says Sam Odukoya, Senior Engineer at CoreWeave.
Lessons for agentic AI transformation
CoreWeave’s rapid and successful adoption demonstrates that agentic AI transformation doesn’t need to be a lengthy, disruptive upheaval. Using North, CoreWeave redesigned a high-value process within a single business quarter, achieving measurable, rapid results.
Rather than starting with a vague goal of “using AI” for customer support, CoreWeave identified specific bottlenecks and inefficiencies in its current system during a workshop with Cohere. This clarity helped CoreWeave to reach its goal quickly. Choosing North enabled CoreWeave to minimize organizational disruption by continuing to work with their preferred, familiar platforms. The change management and internal training to adapt to the new support system were completed within a single business week.
By proving a high-impact use case, CoreWeave has laid a foundation for further agentic AI innovation. Using North, the team plans to deploy AI agents more widely across customer support and critical incident response to accelerate and enrich its responses.
“This is just the beginning,” says Mancuso.
