Software Exists To Do Work For Us

A conversation with Gil Akos, Co-founder and CEO of Astra.

Since the beginning of the COVID-19 pandemic, consumers have been saving more of their money. A lot more:

Morgan Stanley estimates that U.S. households have saved an extra $1.1 trillion as of August, compared to pre-pandemic levels.

And banks, in accordance with the law of supply and demand, have been reducing interest rates on savings accounts:

The average rate paid by banks on basic, federally insured savings accounts — known as the annual percentage yield — was a mere 0.05 percent … according to the Federal Deposit Insurance Corporation.

This, understandably, frustrates consumers. A quick search of “bank lowering my interest rate” in Twitter turns up a litany of amusing tweets like this one:

I think this frustration isn’t so much about the rates themselves as it is about the lack of control that consumers feel as economic conditions change.

It’s like driving. When you’re behind the wheel and you hit a traffic jam, you tell yourself “it happens”. When your Uber driver runs into the exact same situation, you think “this guy’s an idiot”. It’s not exactly fair, but it is how human beings process adversity when the locus of control shifts away from them.

All of which brings us to a question — how do we help consumers feel more in control of their financial lives?

To answer that (and other questions) I spoke with Gil Akos, Co-founder and CEO of Astra about automated money movement and the recent launch of Astra’s API and developer program. Our conversation, edited for length and clarity, is below.

Fintech Takes:

Let’s start with philosophy. You’ve said that Astra’s goal is to “build a platform that makes moving money universal, programmatic, and automatic.” Let’s explore that. Why do you think it’s important for money movement to be automated?

Gil Akos:

Software exists to do work for us. We should be able to say “please, software, relieve the anxiety that I have around my money being in the right place at the right time.”

That’s why we started Astra — we wanted money to move around on our behalf without manual intervention being required on a daily or weekly basis.

And if the rest of our contemporary digital lives (outside of our finances) is any indication of what we expect technology to do for us, the jobs to be done by fintech are extremely underdeveloped. As they say, “everything is fintech” but “fintech is only 1% finished.” We believe automation is the highest leverage technology that we can possibly apply to these challenges.

Fintech Takes:

A while back, I wrote about the 5 Levels of Automated Money and one of the big distinctions between the levels was between ‘closed’ and ‘open’. Closed self-driving money solutions allow you to move money from any account but only into the provider’s account. By contrast, open solutions (like Astra Finance) allow people to move money from and to any account. When you say money movement should be ‘universal’, I think of enabling an open model – where a consumer’s money is able to move wherever it is best for it to go. Is that consistent with your view?

Gil Akos:

Yes, with a slight qualification. ‘Open’ (à la the European banking system) frequently comes with an assumption that all parts of banking are free-flowing and accessible. The way we think about it is more aligned with the self-driving automobile technical framework — it’s about the model being conditionally complete. As you pointed out in your article, it’s the conditions within which the automation works that define how advanced it is as a solution.

As an analogy, Waymo isn’t trying to build a solution where their cars only work in a geofenced area then try to encourage you to use Waymo cars and Google Maps in only those conditions! The magic of Google Maps is FROM any location TO any location – now give me the route and adapt it if you detect traffic ahead. You need both a complete map and the smarts. The same goes for Astra. You should be able to say “send my money from HERE to THERE based on this RULE” wherever those accounts happen to be held. We are working towards making the financial graph fully functional and topologically complete. That’s the ideal: a conditionally complete (ie universal) model.

Fintech Takes:

I also argue in that same piece that there is an important distinction between self-driving money solutions that are powered by consumer-defined rules and solutions that are powered by algorithms that consumers never see or touch. Why do you think a programmatic approach is best?

Gil Akos:

Two key reasons. First, there is just so much core technology to build to make money movement more functional. It’s a complex challenge, and why wouldn’t it be? Card networks and ACH are complicated financial systems. But once we do have the core building blocks in place and accessible through a platform so that moving money is just an API call, fintech companies would be empowered to explore the hands-off algorithm territory more fruitfully. Cloud computing and AWS are the historical parallel. (P.S. Astra is that platform.)

Second, a programmatic approach allows us to map the intent of the user to the rules that power a routine. That’s valuable for the user experience in that it makes the process more directly understandable. I should KNOW when I set up a sweep transfer what should happen. We believe that’s extremely important because the action our automation engine takes should be tangible and not surprising. That’s in direct contrast to a black-box algorithm. The range of what people want to do with their money is broad enough that there is a ton of territory to cover before even starting work on a hands-off solution.

Fintech Takes:

Your point about leveraging users’ intent to help make the experience understandable and not surprising is really interesting to me. There’s an emotional component to financial services that I think is really important — people want to know where their money is at all times and surprises, even ones where the end result is positive, can be counterproductive right?

Gil Akos:

Sometimes the conversation around self driving money gets so focused on the internal machinations of how it could work and what it would mean in an idealized world that we forget to think about what it means in the real world.

When you open up Google Maps, does it just automatically say “I'm going to take you to Starbucks now”. No! You still have to type in Starbucks.

It starts with the user’s intent.

Fintech Takes:

You recently launched Astra’s API and Developer Program. Can you give me the quick (virtual) elevator pitch? What is it and why are you launching it?

Gil Akos:

Astra’s API enables developers to add money movement and financial automation to their products through a simple integration. And it’s all powered by our transfer orchestration system that is extremely reliable and default fast for settlement. 

If you’re a fintech app wanting fast inbound transfers to fund new accounts, a debit card product that wants to ensure your customers’ cards always have available funds for spend, or a challenger bank trying to help your customers move their direct deposits into their account with you, our platform can help. 

We are launching the developer program because our ambition has always been to make intelligence and automation for your finances available wherever you interact with your money. We’ve processed 50,000 transfers approaching $10m in cumulative volume to date through our system for the users of our app, so we know our technology is scalable. This is the natural next step in the evolution of our platform. 

Fintech Takes:

One of the keys to the success of AWS was that Amazon and its enormous e-commerce business was AWS’s first and best customer. This forced AWS to be good enough — in terms of its scalability, reliability, and performance — for it to be a good choice for other non-Amazon companies.

It seems like Astra took a similar approach in building a B2C app that was the first and best customer of the Astra money movement platform. What did you learn building Astra’s app that you think will differentiate Astra’s platform and API?

Gil Akos:

Three years ago we asked ourselves “how are we going to build this automation engine for money movement and how are we going to prove that it works?” And the answer that we landed on was that we would be customer zero and we would run the biggest, baddest beta test that you’ve ever seen by building an app that consumes the API and leverages the automation engine.

Apparently it's not as common as you'd expect in the fintech infrastructure world, but we thought it made perfect sense. In retrospect, it was absolutely essential for us to roll up our sleeves, get into the weeds, and really figure out how this platform should work.

And the biggest thing we learned from that experience was how to create a high performance system that is able to land funds with a degree of reliability that you can't really get anywhere else.

Today, approximately 15% of all ACH transactions fail. That’s simply unacceptable if you’re a consumer-facing fintech company trying to consistently deliver value, through money movement, to your customers.

Transfers between banks in Astra’s system land as expected 99.999% of the time, which we certainly wouldn’t have been able to achieve if we had started by building an API for an assumed customer.

Fintech Takes:

The definition of a ‘primary bank’ or ‘primary account’ has become a lot fuzzier over the last couple of years. Digital account opening lowers the costs to experiment with new accounts — from challenger banks, P2P apps, roboadvisors, merchant payment apps, etc. — without necessarily closing their existing accounts or moving their direct deposits. In general, I think this is a positive trend, but I do worry that it is creating a more fragmented and difficult-to-manage ‘app ecosystem’ for consumers. What role do you see Astra’s platform playing in this environment? 

Gil Akos:

That is spot on. The individual application or bank account may have gotten better and there is more choice for the consumer, but an individual’s system of finance has become much more complex to manage. We call that the ‘financial graph’. The number of transfer flows between those accounts increases exponentially (n*(n-1))! There must be a countervailing force to help us manage that complexity. Luckily, software is really good at that!

We see Astra as a functional connective tissue to an increasingly complex financial graph. If funds can be pushed to and pulled from any node in that graph, and those flows can be automated, consumers can have as many accounts they want from the providers they prefer, and those accounts will work well together. 

Fintech Takes:

What’s the one change to the current U.S. financial infrastructure — UK-style open banking regulations, a real-time payments system, etc. — that you think would unlock the most value for consumers?

Gil Akos:

Faster payment settlement if broadly available across financial institutions would definitely be a great improvement for consumers and small businesses. The trend we are most eager to see accelerate is the ability for fintech companies like ours to gain more direct access to the underlying payment pipes. Whether that is through new charter types or the ability to develop direct developer relationships with the Fed, that should drive down costs and spark innovation for financial technology. That is the largest lever for creating value for consumers.

Short Takes

(Sourced from This Week in Fintech)


Alex Johnson