Sage is, in many senses, an unlikely technology company. Founded in a pub in Newcastle in the early 1980s, dedicated to serving smaller businesses (it provides accounting, financial, HR and payroll software) and not hugely flashy as a brand, it has nonetheless grown to be the largest technology company listed on the FTSE 100 — Sage reported £458 million in revenue during its last quarter alone — and has built up a team of over 80 in-house machine learning engineers and data scientists as it modernises its extensive software offering.
While the company might still have proud Newcastle rather than Silicon Valley roots, the man leading Sage’s ambitious innovation efforts (the company has ramped up R&D spending by over 50% in the past three years) is more of a Silicon Valley native: Chief Technology Officer (CTO) Aaron Harris was a founding member of San Jose-based accounting software company Intact, which was bought by Sage for $850 million back in 2017.
Staying on after the acquisition, Harris now heads up technology for the company which has 11,000+ staff in 20 markets and over two million customers (in Britain alone, an astonishing 43% of all businesses pay staff through Sage’s software) and also takes the strategic lead on a growing number of technology acquisitions.
“We are a technology company so [my team] works to make sure that technology strategy and and the company’s purpose and ambition are all highly aligned. What we do with our technology is the biggest part of driving our purpose and our ambition” he tells The Stack, adding: “I have several teams that build and design the enabling technology that will help us to realise the company’s vision and purpose. We’ve got a big AI team, a big data science team, we’ve got an architecture team that’s developing next-gen digital architectures…”
The softly spoken CTO reports directly to CEO Stephen Hare, working closely on the executive leadership team with Chief Product Officer Walid Abu-Hadba; a Microsoft and Oracle veteran who was appointed in May 2021: “He and I are very close partners”, Harris says. “I lead the teams that are developing more emerging initiatives that we’re working on and Walid’s got the bulk of the product teams, so product management engineering.”
Predictive analytics and machine learning are a growing part of Sage’s strategy. The software firm recently bought Israeli SaaS startup TaskSheriff and Ireland’s AutoEntry (which uses AI and optical character recognition to automate data entry) in a bid to further turbocharge its ability to further automate and improve its offering.
But it has also organically grown out its own AI team – Sage keeps the vast majority of IT functions in-house rather than outsourcing them – and as Harris notes, “between AI and ML engineers and data scientists, we’re growing that team to close to the mid-80s this year: it started with around 10 people three years ago, so we’re growing as fast as practically possible”, adding on an April 2022 call with The Stack: “What you don’t do is go out and hire 1,000 data scientists.
“We’re roughly doubling every year, but it’s at a pace that makes sense for us…”
Sage CTO: “We’re huge fans of serverless…”
Like many “traditional” software providers Sage is pivoting slowly but surely from on-premises software to cloud-native SaaS offerings for its customers. Harris notes however that “when a small business or medium-sized business chooses an accounting product, a finance platform, they integrate a lot of the workflows in their business to that accounting product and they don’t change it very often. It’s roughly a 10 to 12 to 15-year cycle for a business to make a decision on their accounting product. So our philosophy is pretty simple. We embrace all of our customers, whether they’re on our connected products that are deployed on-premise, or cloud customers, which, you know, frankly does get the bulk of our investment. But there’s no forced migrations. There’s no programmes in place to force a decision on these companies. Our intention is to great to create the best possible migration path for them. When they see the value… we want to have great programmes in place for them to move to our cloud native products.”
Sage’s own internal IT is also evolving as the company grows.
“The bulk of our product development is in public cloud architecture” the Sage CTO says.
“We make meaningful investments in both AWS, Azure, and to a lesser extent, Google. We try to strike a balance of investing for scale efficiency. So in other words, we don’t want 14 different relational databases out there, but we also believe that cost is less of a constraint in the era that we’re in – and in a lot of ways, we need to balance for agility. So we’ve got a couple of graph databases that we use; we’ve got an append-only ledger database that we use… we’re not overly focused on standardising on any particular programming language.
“We want our teams to use the language that’s right for the problem. So Python is usually the solution for AI work; .NET core is a big part of the stack we use for building our SaaS applications… but the overriding theme is [to] embrace an architecture of services, of agility, of interoperability: we’re huge advocates of serverless computing… The actual underlying implementation is less important” he emphasises, painting a broad brush picture for us quickly: “We’re a big company; 11,000 colleagues… We’re on Teams today because that’s our corporate communication platform. We’re moving more and more to a “no trust” approach to technology… our colleagues don’t log into a VPN and then access a castle of shared services; each service is deployed to the cloud, it’s independently protected. A few services are standardised on Salesforce for CRM products. But broadly we are evolving to a cloud services-first approach with a no trust cybersecurity approach…”
That’s all just a delivery mechanism really for what his team is building and he’s clearly far keener to talk about the innovation shop floor, as it were, than the BAU machinery, saying: “When I when I joined Sage, I’d competed with it for 17 years… Sage is a bit unique in being as focused as it is on accountants and small business owners and CFOs. So it’s really respected even if over the years it hasn’t had the most advanced technology.
“When I joined and especially as I took on the CTO role what I realised was that there’s actually a lot of really exciting stuff happening in Sage with technology” he says — pointing to the emergence of Open Finance/data and APIs in “accelerating the digitisation of standardised data flow” before adding apologetically: “That sounds like techno jargon babble, but it’s got an amazing ability to drive efficiencies and markets…”
Operating with purpose
Hence the growing team of data specialists — a competitive area in which to hire people right now.
Asked if he found recruitment of data scientists and engineersa challenge Harris says that those that go into data science and AI are “really attracted to a purpose. They’re really attracted to this belief that they can do something that’s going to have an impact… we’re at a great stage where we’re growing fast; they’ve got real impact; we’ve got a clear vision we can communicate. [For example] In the UK small businesses missed out on about £25 billion in credit… We really do believe that not only can we improve that access to credit universally, but that there are portions of our communities that are disproportionately disadvantaged by this. We know black-owned businesses get paid a lot slower than other businesses; that there’s real tangible gaps between the rate at which African Americans qualify for mortgages with the same credit history as a white applicant. A very powerful approach to fixing some of this inequity is to digitise those processes; to keep human bias out of a lot of decision making, and to create data standards that… have a real meaningful effect on levelling the markets.” [nb: Hyperlinks by The Stack]
“So that’s what my data scientists are working on” he says, adding: “We’ve got three areas of focus with with AI.
“The first area of focus is in automation; eliminating the financial close. The accounting industry works on monthly closes, quarterly closes, annual audits and there’s all this repetitive work that happens to get through those cycles. With AI, we can create automated services that fully automate all of the processes within the accounting team: so you’ve always got real-time data, everything’s automated; there’s no close anymore because you’re always current. That’s one big area of focus. The next big area of focus is a co-equal investment in trust and assurance.
“We’ve got some really exciting AI products that essentially build patterns of understanding in our customers business and [which] in real time identify when transactions flow through the business that are anomalous to those patterns, so when when a finance team goes to approve a transaction, we have AI that supports the review of that human by highlighting where are there anomalies in this transaction. [This] results in a huge increase in confidence, especially as you go into medium-sized businesses where they’re dealing with potentially tens of thousands, or hundreds of thousands of transactions a month. Humans don’t have the capacity to review all of that.
“The traditional approach in technology is to create business rules and configuration and processes that control the way things happen. You can’t possibly cover for everything. That’s where AI comes in…
“The third area of investment is… if we can automate the routine tasks of assurance and the accounting workflows that frees up the accounting staff to focus on more strategic activities. We can apply data science and AI to look into the future: so real time insights, AI-powered forecasting: there’s obviously lots of opportunities there to to help small businesses with cash flow forecasting, to help medium-sized businesses and specific industries to manage and look at how renewals are going to come in,” he says, noting that “one of the things that we learned pretty early with customers on this is that the real value in in AI powered forecasting isn’t so much that it’s more accurate than humans; the real power is that AI can do it continuously, and basically for free.”
That means that companies that might previously have only been able to forecast renewals quarterly can now get trend-spotting in real-time, he says and flag material changes in forecasts.”
Sage’s ability to really double down on value-added services for customers powered by data is going to depend in no small part on, as Harris noted earlier, the continued evolution of “Open Finance” — the business version of the Open Banking that has inspired so much activity on the consumer banking side.
In the UK the Department for Digital, Culture, Media, and Sport (DCMS) in November 2021 published a policy paper, snappily titled National Data Strategy Mission 1 Policy Framework: Unlocking the value of data across the economy that emphasised “the need for the government to improve coordination, incentivise data-driven innovation and address perceived risks to data sharing” — there’s a lot of work to do but the direction of policy travel, no matter how slow, is clearly towards supporting a more open data environment across verticals and in a challenging market for SME finance, Sage may yet find itself an increasingly pivotal player in an emerging Open Finance world.
Sage, meanwhile, says it develops its AI tools alongside customers that opt-in to data sharing, with Harris noting: “It could be 10 customers [it] could be as many as 100 customers [that] join a programme where they share data with our data science team — obviously cleansed of personal information. We use that data to design our machine learning models; once those models are [ready] we deploy them to production for all of our customers. The one thing that we just can never break is that trust because it’s such an incredibly valuable asset that we have access to, but we don’t own it — we can only use it if our customers trust us to use it in the way that works for them.”
Follow The Stack on LinkedIn