A talk with ThoughtMachine on VaultOS and the future of core banking

Core banking is considered for many as the gordian knot in the digitization of banking. As a result, a vast majority of fintech companies are innovating around the core, leaving legacy IT to the banks.

However, some entrepreneurs are audacious enough to attempt to renew core banking from scratch. I had the pleasure of meeting Tristan Fletcher of ThoughtMachine in London earlier this fall, and used the occasion to ask him a couple of questions of which problems ThoughtMachine is attempting to solve with VaultOS.

 Everyone is complaining about legacy IT, but few actually spell out what is wrong with today’s core banking solutions. Which problems are Vault OS going to solve, and how does the system differ from legacy systems in terms of architecture?

New product development often suffers from inertia in the sense that every new product adds additional complexity to an already complex core. This also limits scalability in terms of number of customers and/or products. Vault OS intends to enable new product offerings with a cost that is independent of complexity, as well as providing an infrastructure that scales in terms of number of users and product offerings with minimal incremental cost.

Risk and compliance is a cost driver for every bank out there, and manual compliance reporting drives both cost as well as increasing risk due to long lead times in terms of data expiration dates. One of the key components in Vault OS is real time treasury management and reporting. In addition, cryptography and security is baked in from the bottom up, and is not just an overlaid afterthought.

Architecture in legacy IT is often prone to repeated expensive and reputation-damaging failures. Vault OS micro-services architecture is built out of many independent components and services surfaced through their own APIs. When one component / service goes down, it doesn’t necessarily impact the rest of the system. Furthermore, Vault OS’ integrated health monitoring processes recognise if a component is down and restarts it. In addition to this, Vault OS aims for a separation and genericization of products running on a common data source, the transactions ledger. This is done by leveraging a Blockchain-style centralised, permissioned cryptographic ledger for storing transactions, products (smart contracts) and actions.

Open banking is one of the most hyped stories in banking these days, how is Vault OS designed in relation to this scenario?

The federated permissioning that blockchain style technology enables is ideal for interfacing with open banks.

How do you plan to utilize distributed ledger technology as an internal part of the system?

All transactions are stored on the ledger, with each transaction being signed by the bank and one or more counterparties, and then a hashing function applied to it and the one above it in the ledger to check for integrity. Secondly all products are encoded as smart contracts and these themselves are similarly recorded in an immutable ledger. Lastly all actions that a member of staff in the bank perform, from the CFO to the cash teller, are recorded in a third ledger.

In what way will Machine Learning benefit banks running ThoughtMachine’s core banking solution?

Vault OS use machine learning techniques to forecast how the customer base will evolve as well as how economic conditions such as a changing yield curve or exchange rates might impact them to carry out sophisticated scenario planning and balance sheet forecasting. This will counteract todays myopia of consequences to balance sheet health of different economic scenarios. The single source of truth that Vault OS’ ledgers embody, enables a more holistic look at fraud in the bank based on a wide range of data: not just card fraud or payment fraud.

When it comes to everyday banking Vault OS also use machine learning to categorise every user’s transactions so that they can see where their money goes (for example x% on shopping, y% on bills) as well as being able to associate transactions with merchants where relevant. This also enables the system to predict how much a user will spend in each category going into the future.

Cloud is unfortunately still frowned upon by the FSA in some countries. From your perspective, what are the benefits of a cloud based core banking system?

A cloud based approach offers no lock-in to specific hardware which results in cheaper recruitment as well as maintenance. This also gives increased reliability, reduced friction for frequent deployments as well as built-in load balancing. A cloud-based approach also provides horizontal scalability and disaster recovery built in from the start, allowing IT costs themselves to scale with demand.

You told me earlier that ThoughtMachine is built in a modular way, allowing banks to choose any modules they want. Are these fully independent of each other or are there some core components that are necessary to tie it all together?

The system “kernel” which is essentially products, transactions, auth is required. As the whole system is API-based the rest can be switched on or off. Vault OS’ workflow system also enables flexibility with regards to external services and business processes are flexible.

Finally, when will we see the first banks taking the leap of faith and doing a core banking renewal with ThoughtMachine?

We are in the advanced stages of trials we are conducting with a variety of banks.

Image: ThoughtMachine