Now Accounting Goes Invisible – Sage Talks AI and Machine Learning

The rivalries between accounting software companies are intense but at least they all agree on one thing. They all want to knock off the mechanical aspects of compliance.

Intuit has a vision called “accounting is done”, Xero says it is “no-code accounting”, and Sage has joined in with “invisible accounting”.

None of them have come out and said “tax accounting is dead”, so that debate will continue to rage. At least until the collective vision is realised and we find out what those slogans mean in practice.

Sage has put some starpower into its mission with the hiring in February last year of Kriti Sharma. Sage’s VP of bots and artificial intelligence is something of a wunderkind.

Sharma graduated in engineering and has a masters in advanced computer science, and at the age of 21 was elected a Rajiv Gandhi Science Fellow for her work in energy optimisation and its applications in astrophysics, material science, polymer and bioinformatics research.

When she was 26 she worked for Barclays Bank as VP, head of product, real-time big data analytics.

Sharma is also the founder of Messaging Bots London, a community of more than 1000 bot developers in Europe. Soon after joining Sage in February last year Sharma launched Pegg, which Sage claimed was the world’s first personal chatbot for business finance.

I recently spoke to the 29-year-old about how she was going to make “accounting as easy as texting”.

Sholto: So what’s the idea behind Pegg?

Kriti: The idea behind creating a conversational interface for business finance products is to make it as easy as talking to a friend. Accounting is built on conversation. Business owners go see the accountant face to face. That’s how it all started decades ago.

Then we got on to the online world and we tried to make it more efficient, more real time and reliable. But storing the data was still quite expensive and AI was not as mature so everyone got the same experience.

Now in the past two years the cost of storing data is much cheaper thanks to the cloud. We can process information much faster, find the next best action (insights), and use natural language communication. That is a game changer. We are getting closer to normal human behaviour as possible.

We are also spending so much time in messaging apps. We can have more one-to-one conversations, that’s what humans prefer these days.

Sholto: How did you come up with the idea for Pegg?

Kriti: There’s a huge opportunity to disrupt accounting and make it as easy as chatting. That’s where Pegg got into the picture. Accounting has become too complex, you have to be an accountant or tax specialist to use the software. But now freelancers and startups and entrepreneurs and a lot more casual business users are using the software as well.

Sholto: So what can Pegg do?

Kriti: Pegg is for the 75 percent of business owners who don’t use any software, for accounting,they just use pen and paper. Pegg can do admin for you and answer questions such as how much you spent on specific things, such as travel. It can tell you who owes you money. It does invoicing, can give you deeper insights, set alerts and reminders.

Now we are taking it to the next level by adding a couple of key things. We are integrated Pegg with Sage products. Pegg is already integrated with Sage One in a couple of markets, and more countries are coming this year. (Only UK and US, with South Africa and Australia to follow.)

We had a huge amount of interest from customers using our enterprise products, Sage 50 and Sage 200. We are also integrating it with the accountant’s version, Sage Accountant Cloud.

We have also previewed a voice capability as well. Voice is challenging because you have to be careful about how you use it. You don’t want fake business owners to ask questions about profitability. You need to design with security and privacy in mind.

Sholto: What are the security concerns?

Kriti: You  want to put in additional layers to be absolutely sure you’re talking to the right person. Google Home has the same challenges. If you just want the balance that’s ok, but you can’t ask a second layer of questions without a pin code or voice matching. Otherwise anyone could ask. There’s a huge user education piece involved.

Sholto: What do you get in the free version of Pegg versus the paid version?

Kriti: Right now we just have the free version of Pegg. Most basic features are task based. For example chasing payments, creating invoices and expenses. You can also look up your top customers, how is my revenue looking, and what are we spending money on. That has been quite widely used.

Beyond that interface is how is my business doing compared to others (i.e. benchmarking). IT will also predict cashflow and integrate with HR and payroll solutions. We have previewed voice integrations in the US and the UK but you need to think about use cases. You can’t just upload a photo of a receipt by voice. So there are some use cases that are different.

Sholto: Leaving aside chatbots, where are things with automation of bank rec? Are you also looking at AI-driven approaches to coding transactions?

Kriti: All vendors have to do auto-coding. Using machine learning on a transaction level, for categorisation, is pretty standard. We are using machine learning for benchmarking so you can compare yourself to others.

Sage recently acquired Compass, a Silicon Valley company, to bring collective intelligence and machine learning much faster our customers. I strongly believe that’s where the future is. You have to do tasks faster and faster. If we can create an assistant with AI at scale then it automates a lot of the tasks.

Sholto: Do you have a timeframe for adding auto-coding to SAge One and Sage Live?

Kriti: It will be sometime this year.

Sholto: What about your API strategy for connecting to other data sources, such as the finance industry?

Kriti: That’s a huge priority. We have announced the banking cloud (bank feed integration) with key partnerships with a bunch of banks. We recently built and launched the banking cloud bot that automates the process of pulling transactional data for the matching. For bank feeds, and capturing the data from receipts. We will be talking more openly about bank feeds.

Sholto: A bot doing bank feeds? Is that the same thing as, well, normal bank feeds?

Kriti: There’s an automation bot that automates the process of taking the data from the bank feed and categorising it. An algorithm has one component to it but a bot is end-to-end automation; of pulling the data and using the algorithm and deploying it in a user friendly manner.

Sholto: Is anything else happening on the API front?

Kriti: We are very much taking a platform approach. Sage integration cloud allows partners to very quickly access the ecosystem. Sage Live is of course is much more powerful with a huge number of apps.

We are also getting new partners on board. For example integrations with navigation, can we get more real-time or manage a fleet or use GPS data.

Sholto: So in short there’s a lot going on. What’s your main focus?

Kriti: We believe voice is a huge frontier. We want to get users accounting-related information and insights in a very simple way. We are automating as many admin tasks as we can and finding smarter insights in the data. And product integrations, especially expanding into HR.