Sedai
How Sedai is Finally Solving Cloud Cost Optimization

For years, engineering teams have struggled to fight back against rising cloud costs. But with countless resources that need attention, from AWS to Azure to Google Cloud, engineers just can’t keep up. The result is millions of dollars wasted, basically flushed down the drain.

This was the problem that inspired Suresh Mathew to found Sedai.

Today, Fortune 500 companies deploy Sedai’s platform to handle cloud cost optimization completely autonomously, without any manual effort. The platform has already performed over 25 million actions to optimize cloud resources, without ever causing an issue in production.

So how did Sedai finally crack the code of cloud cost optimization? And what makes it safe?

“Don’t make your engineers optimize cloud costs. This is a job that AI can do safely, right now,” says Mathew, founder and CEO.

The Need for an Autonomous Cloud

Cloud optimization looks simple, on the surface. For example, when an application requests more CPUs than it appears to use, the obvious solution is to scale down the CPUs to that level.

Yet experienced engineers know that looks can be deceiving. This new configuration could degrade the app’s performance or even cause an outage, resulting in customer dissatisfaction. According to Mathew, there are too many variables for the manual approach to work; from internal changes, such as new feature releases, to external changes, like increased traffic. That’s why site reliability engineers (SRE) often over-provision resources on purpose.

But with Sedai, companies no longer need to decide between cost and risk.

Don’t make your engineers optimize cloud costs. This is a job that AI can do safely, right now.

“I spent a lot of my engineering career doing optimizations, so I know from experience that manual tuning is extremely boring and extremely risky,” Mathew said. “We built Sedai to evaluate every signal and metric from the cloud, every time it takes an action. To handle the complexity of the real world, you need to go autonomous.”

Safety through Reinforcement Learning

Built on patented machine learning models, Sedai’s platform continuously balances cost, performance and availability in real-time. This enables Sedai to perform intelligent optimizations that reduce cloud costs by as much as 50 percent, while also enhancing the safety and reliability of its customers’ environments.

We asked Mathew how Sedai differs from conventional “auto-scaling” tools.

Across the board, these tools rely on some combination of automated usage rules or fixed performance thresholds. Sedai is unique in that it understands and adapts to the behavior of each application. As a result, the platform can distinguish between a CPU spike during a batch job, such as, versus one during peak user traffic, enabling it to take the correct action autonomously. The company has been awarded eight US patents, primarily for this safe-by-design approach to cloud optimization.

“No two clouds behave the same way, especially at the massive scale of a modern enterprise,” said Mathew. “The key for Sedai is deeply understanding each environment so that it can find the perfect blend of cost and performance, based on each customer’s SLOs. We’re very proud that only Sedai is patented to make safe optimizations.”

Autopilot at the Highest Stakes

Sedai has earned the trust of some of the world’s most security-conscious organizations. Palo Alto Networks, for example, needed to maintain 99.999 percent uptime for its cloud, given the high stakes of cybersecurity. Understandably, engineering leaders tend to be skeptical of letting autonomous systems run in production. So Sedai started by making recommendations to prove its safety, before shifting to full Autopilot.

The impact was a $3.5 million reduction in Palo Alto Networks’ pre-discount cloud costs , while also saving more than 1,000 hours of toil for the company’s busy SREs.

“I love seeing a customer switch to Autopilot when they realize it’s the safest way to manage their cloud,” said Mathew. “For me, it’s not just about Sedai’s success. It’s about talented engineers getting to finally focus on building new products.”