Interview

AITech Interview with Aran Khanna, CEO and Co-founder, Archera.ai

Archera.ai

Aran Khanna gives a sneak peek into the recent developments at Iterative.ai and highlights the most exciting trends in AITech right now.

1. Tell us a bit about yourself. What drove you to start Archera?

I was born and raised in Seattle, and my entire upbringing was around technology. Both of my parents worked at Amazon and Microsoft in the early days, and my first tech job was at Azure. So you could say I have come full circle.

Later, in college, I worked as a researcher with the former FTC CTO, Latanya Sweeney, on various privacy and technology related projects. I then moved on to researching ML, working on the  Deep Learning framework MXNet, which was adopted by Amazon, leading to me joining AWS.

I had the idea for Archera during my time as an engineer at AWS. I was part of the original Sagemaker team and was tasked with driving customer adoption for its very expensive GPU compute product. Essentially, all customers had difficulty making comparisons. They didn’t know what they’d be paying for services, and many were too uneasy about predicting usage to commit to services for a year-plus in order to get a discount. Visibility into who spent what was getting better, but using that to drive predictability and then optimize for usage and costs in a low-risk way was out of reach for most engineering teams. This goal was a big part of the inspiration for Archera and also a large motivation for the emergence of FinOps—which some might call IT finance—a new practice area dedicated to cost-optimizing cloud contracts, commitments and service deployments, against technical & business requirements. 

2. What challenge do you solve for your customers?

Archera helps customers make their cloud FinOps practice uncomplicated, automatic and risk-free, whether on a single major cloud provider or multiple.

We start by providing advanced cost visibility for free, to help customers understand the drivers of their expenditures on cloud services. There are several pieces to that. Attributing costs to the right team or department confounds companies, with many systems and developers running on a single AWS organization or common set of accounts—Archera makes it straightforward to carry out that otherwise very difficult task. 

Archera’s true value comes in providing visibility into costs. We help customers build detailed usage forecasts and automation policies that can greatly optimize cost, minimize risk, and drive increased predictability—and on top of this Archera provides insurance against commitments for customers, arranging to buy back capacity they pay for, but then do not use. That ensures they are never over-committed and they are free to change their commitment strategy when it is most convenient, not just when contracts expire. It means they do not leave money on the table.

3. Who is your ideal customer? Give us an example?

A great customer for us is any business that does not want to waste large amounts of engineering and accounting time figuring out cloud services and costs. Archers is also a good fit for companies that suspect they are spending more money than necessary because real-time management of cloud contracts and resources is so complex. Unfortunately, that suspicion is probably well-founded; many organizations overpay and pay for resources they won’t use.

To be more specific, though, our ideal customer is probably growing fast and has a sizable public cloud presence, handling migrations to the cloud or between cloud services, with seasonality or unpredictability in their usage. In other words, customers with large and variable AWS usage will benefit greatly from Archera.

They are organizations that are eager to optimize, and ready to put effort into getting it right. They want to reduce risk, and they see the benefit of our insurance, where we guarantee to buy back unused commitments. The buyback insurance opens up their choices; it allows them to become much more aggressive in seeking discounts and covering short-term, uncertain usage.

It’s even better if they have someone who really understands engineering needs, business constraints, complex contracts and cloud costs—a person who has collected all that in one brain —because that person will see how to take full advantage of Archera. You can call them IT-Fin or FinOps or Engineering Finance or DevFinOps—whatever the title, they are valuable to their employer.

4. What ultimately needs to happen to help companies manage their cloud costs?

While cost visibility, attribution and chargeback are important, what is really needed as cloud vendors shift to per-second billing models for all services is a common way to forecast, compare and plan usage/cost-outlays between cloud providers, as well as the many managed services within them. The next step is to deploy intelligent automation to optimize in real time against those forecasted plans. This is a big shift in the operating model that requires automation, focused software and machine learning to really get right.

5. Is this a problem the big clouds should solve? Or should there be a standard?

Standardized billing is long overdue. Everybody understands that the billing methods of the Big 3 are a nightmare to untangle and compare. There is too much at stake, and the numbers are too big for this to continue. Conversely, the profitability is too big for the ‘Big Clouds’ to stop it. Even with good intentions, I would not expect Google, Amazon and Microsoft to provide ways to cut the total number of dollars committed to their platforms. There has to be a neutral third party to optimize cloud usage and commitments for the customer. That’s where Archera comes in; we automate, insure, aggregate and resell unused commitments to help customers navigate the complex pricing environment.

Just to sum up, non-standardized billing creates several sets of problems, from tracking expenses and attributing them to comparing and managing commitments across multiple vendors.

6. Has the pandemic driven more cloud adoption and therefore cost and complexity?

Yes. The forced explosion of remote work and the growing acceptance of using many special-purpose SaaS apps to be nimble and succeed in digital transformation has driven cloud adoption since the start of 2020. Organizations became more dependent on cloud computing, with remote work accentuating that dependency in 2020 and 2021.

There’s a report from Gartner saying companies that are unaware of mistakes in their cloud choices will overspend by 20 percent to 50 percent. That just validates the benefit we can offer to nearly any company running parts of the business in the public cloud.

7. Could AI play a role in helping solve this challenge?

Yes, and we built AI in from Day One for specific purposes—for example, to predict usage, price insurance and make a market between customers for unused cloud commitments.

AI helps in making better forecasts, and also to reason more accurately about future uncertainty.
I credit AI with allowing Archera to put skin into the game and take on risk on behalf of customers. We are an AI-first platform, with modeling uncertainty and risk management at the core, and that’s partly a result of my background as a researcher in machine learning as well as our CTOs background in quantitative finance. For us, AI is a key to enabling great forecasts and insurance models. That includes projecting an organization’s future cloud usage to help them plan effectively. It can also help identify cost spikes faster than even a very attentive manager can spot them, and find the root cause.

8. How do you envision this space looking in 5-10 years? Will this problem be solved? Or will companies continue to be challenged by cloud billing?

In the near term, complexity will prevail. You have universal adoption of services ranging from ML to databases to commodity compute from the three biggest cloud providers (AWS/Azure/GCP).  Into this picture, you add the third-party managed services like Snowflake and Datadog, which take individual slices such as data warehousing or monitoring. They do a very good job in their specific markets, too. In the longer term, users can expect to see in cloud cost management something like what Datadog did with monitoring—whether Archera or someone else provides it. We envision this as a platform sitting across all major vendors that can roll them all up, with cost metrics and monitoring, billing and contract optimization. It’s a tremendously complicated problem that is still in the early stages of being conquered.

We may also see some tech companies save a bundle by repatriating some operations from the cloud in the near term, but in the long term the leverage of the multi-tenant operating model will win out.

9. What advice do you have for companies going multi-cloud now?

The tough thing is in the near term due to the complexity you’ll want to hire an expert for each major cloud platform you adopt, and that’s very difficult given the hiring environment. In fact, it won’t give the best outcome. Simplify where you can, starting with your “cloud-zoning policy”, have major production services stay within a single provider where you are very confident in your team’s ability to operate. You probably can’t get all the specialists you’d want, so figure out what is the Minimum Viable Skillset you can work with.

Some broader advice for teams is to get the “cloud-zoning policy” in place first, make sure overall it’s coherent, know where everything will live up front, and implement the policies to enforce that resources are being deployed in the correct provider.

10. If you had to go back in time and give yourself one piece of advice regarding your market, what would it be?

I’d tell myself, “simplify everything to the extreme.” It’s already so complex, we can only get customers to understand the opportunity if we make it 10x simpler. I talk to super smart executives all the time and the problem we solve is way over most of their heads. They know the cloud vendors make it incredibly complicated, and the core of Archera’s success is that we make it simpler and clear for novice and advanced users alike to supercharge their cloud FinOps strategy.

Visit AITechPark for cutting-edge Tech Trends around AI, ML, Cybersecurity, along with AITech News, and timely updates from industry professionals!

Aran Khanna

CEO and Co-founder, Archera.ai

Aran Khanna is the CEO and co-founder of Archera.ai, a Cloud Resource Automation platform that enables organizations to operate dynamically in the cloud. A Harvard Computer Science alum & former Amazon product, Aran is focused on understanding the consequences of the increasing role of technology in our lives, particularly in the realm of personal privacy and impact of machines learning systems. Aran has given many talks on the consequences of a digitizing society including a popular TedX talk . He has also published widely cited research in the deep learning field in collaboration with CalTech Professor Anima Anandkumar, as well as multiple papers on consumer privacy with former FTC Chief Technology Officer and Professor at Harvard, Latanya Sweeney

Related posts

AiTechPark Interview with Balaji Sreenivasan, CEO, Aurigo Software

AI TechPark

Interview with Steve Shillingford, President and CEO of DeepSee

AI TechPark

Interview with Justine Fox, Director of Software Engineering at NuData Security, A Mastercard Company

AI TechPark