What's Happening?
Together AI, a company focused on open-source AI infrastructure, has implemented Serval's platform to automate 95% of its just-in-time access requests. This partnership aims to establish a robust, transparent, and scalable security foundation for Together AI, which
has experienced rapid growth and increasing complexity in its internal systems. Previously, Together AI's IT ticketing and access management were handled manually through Linear, a system not designed for IT use cases. This led to operational inefficiencies and heightened security risks as the company scaled. The security team, led by Derek and Todd, along with IT support specialist Kate, sought a more flexible and automated solution after evaluating other options like Zendesk and Freshservice, which were deemed insufficient for their complex identity workflows. Serval's natural language interface proved to be a key factor in their decision, allowing for the rapid creation and publication of workflows.
Why It's Important?
This collaboration is significant for the technology sector, particularly for fast-growing AI companies. By automating 95% of access requests, Together AI can significantly reduce the time and resources spent on manual access provisioning, which was previously error-prone and time-consuming. This efficiency gain allows their lean IT and security teams to focus on strategic projects rather than day-to-day operational bottlenecks. From a security perspective, Serval's automated deprovisioning ensures that access is removed promptly, enhancing compliance and overall security posture. This approach provides necessary guardrails and visibility without hindering employee productivity. The success of this implementation at Together AI could set a precedent for other rapidly scaling tech companies looking to optimize their security and IT operations, demonstrating how advanced automation can support growth while mitigating risks.
What's Next?
Together AI plans to expand Serval's automation capabilities further into various IT functions. This includes automating onboarding and offboarding workflows, device and SaaS application provisioning, and common IT requests such as password resets and distribution list changes. On the infrastructure side, the company aims to automate more Terraform and AWS IAM role workflows, transitioning from manual processes to policy-driven automations. With the day-to-day operational tasks handled by Serval, the IT and security teams will shift their focus to higher-value initiatives, such as developing a more mature identity governance program, redesigning the employee onboarding experience for greater efficiency, and investing in strategic security architecture improvements. Other teams within Together AI, including engineering and People Ops, are also exploring Serval for their own workflows, indicating a potential broader adoption of the platform across the organization.
Beyond the Headlines
The adoption of Serval by Together AI highlights a broader trend in the tech industry towards intelligent automation to manage complex IT and security landscapes. This move reflects a strategic shift from reactive problem-solving to proactive, policy-driven security and operational management. The 'human-like' natural language interface of Serval, which users often mistake for interacting with a person, points to the increasing sophistication of AI in user experience and its potential to seamlessly integrate into daily operations. This could lead to a redefinition of roles within IT and security teams, moving them from execution-focused tasks to more strategic and innovative endeavors. Furthermore, the ability to build complex workflows in hours rather than months underscores the transformative potential of such platforms in accelerating business processes and fostering agility in dynamic environments. The ethical implications of AI-driven automation in access control, particularly regarding transparency and accountability, will also become increasingly important as these systems become more prevalent.













