With the global cloud FinOps market expected to grow from $832.2 million in 2023 to $2,750.5 million by 2028, with an average annual growth rate of 18.8%, companies offering data observability and FinOps have found rising prevalence in the market. One such player who has been in the market for over a decade, leveraging machine learning and automation to provide services to some of the biggest cloud players such as Databricks, Snowflake, Big Query and others is Unravel Data.
AI is Not New
Using AI-powered tools and an insights engine, this US-based company which also has an office in Bengaluru, has been relying on in-house ML models and algorithms. “AI is not new to Unravel. We have used AI to automate tasks for data teams over the last 10 years after observing more than 50 million data pipelines and queries. Today, AI is woven into Unravel’s platform at all levels,” said Kunal Agarwal, CEO and co-founder of Unravel Data, in an exclusive interaction with AIM.
Unravel Data’s ML algorithms have been developed in-house and have been trained across a wide variety of workloads for each specific platform to ensure maximum accuracy in insights and predictions. The company’s AI-powered Insights Engine utilises a robust tech stack that starts with data collection from diverse sources, covering big data application performance, cloud expenses and historical usage patterns.
While AI has been the core of Unravel’s business functioning, generative AI is not far behind. Speaking about its implementation, Agarwal mentioned that Unravel has big plans with it and will be shared with the market very soon.
Customer-Centric Solutions
Increasing adoption of data engineering teams that are guided by DataOps practices will lead to fruitful results. By 2025, data teams supported by dataops tools and practices are said to be 10 times more productive than teams that don’t use Data Ops. With the need to stand apart and offer specialised solutions, Unravel has addressed that as well.
“The AI isn’t just reactive; it employs predictive analytics, forecasting future cloud spending based on historical data and trends. This foresight empowers businesses to make proactive adjustments, avoiding budgetary pitfalls. This also means that our ML models are trained for each specific platform, across a wide variety of workloads to provide accurate insights,” said Agarwal.
Catering to specific needs, Unravel has distinguished products for each of their big customers including Databricks, AWS’ EMR and others. Unravel’s purpose-built AI provides insights in real time at the job, user, and workgroup levels to help teams improve their cost allocation and workload efficiency. Furthermore, a standout feature of Unravel’s Insights Engine is its ability to act as a financial detective. It scrutinises cloud spending patterns, identifying anomalies and inefficiencies in resource allocation. “This is invaluable for organisations aiming to streamline costs and enhance operational effectiveness,” said Agarwal.
Agarwal believes that the AI-driven resource rightsizing recommendations are akin to having a personal trainer for one’s cloud resources. “Unravel Data ensures that your resources are neither underutilised nor oversized, optimising costs with precision. The AI also plays a crucial role in cost allocation, accurately attributing cloud costs to different business units or projects.”
Data Observability in 2024
With booming predictions for cloud end-user spending which was expected to hit $600 billion in 2023 as per Gartner, the forecast is only going to hit higher, thereby spiking the need for data observability and FinOps platform.
“In 2024 (and beyond) cloud data costs are going to be much higher because you’re gathering, retaining, and processing more data. Data observability to understand what’s going on with data applications/pipelines will become table stakes. What companies will really need are solutions that leverage data observability with FinOps and AI-powered recommendations that optimise performance and costs of data workloads,” said Agarwal. However, challenges of navigating around generative AI in FinOps will continue.
“The total impact, both fiscal and environmental, will have companies putting greater scrutiny on their AI projects, such as which models do they really need to run, which projects needs generative AI, can a model be repurposed/fine-tuned as opposed to starting from scratch, and rightsizing jobs to ensure that they’re wasting neither resources nor money,” said Agarwal, who believes these are some of the nuances that needs to be dealt with.
Companies such as Dynatrace, Datadog, Microsoft System Centre among others are some of the notable competitors to Unravel Data.
