Save content
Have you found this content useful? Use the button above to save it to your profile.
Drone Point View of City Street Crossing at rush hour.
istock_Wenjie Dong_aweb

Deep APIs: An integration interrogation


Richard Sergeant looks at the more complex capabilities of modern integration and what we can expect from APIs and AI in the years to come.

14th Oct 2020
Save content
Have you found this content useful? Use the button above to save it to your profile.

Integrations, or application programming interface (APIs), are increasingly becoming the norm for accountants and their clients. With protocols for security and quality in place, it’s now rare for a new piece of technology not to integration capabilities.

Primarily, integrations are designed to push and pull information from one system to another.  Receipt capture to the ledger, and then through to the reporting software is a good example of a string of connections which do discrete parts of a longer workflow. 

Visions of technology like artificial intelligence (AI), machine learning and robotic process automation (RPA) are still some way into the future. But with RPA focused on bulk tasks and removing intervention, there are still signs of what a closer working relationship between suppliers will be able to deliver. 

However, we have yet to see what a more connected vision of these systems might look like.

Deeper integrations: FreeAgent and Hiscox

An example of going beyond pushing and pulling to a deeper integration is the current pilot between FreeAgent and the insurer Hiscox.

By scanning the data within the ledger, FreeAgent is able to spot whether transactions for insurance are being made, and work out what might be needed by the business. Complex modelling allows alerts with indicative quotes to be supplied. 

If the client wishes to explore further they are passed to the Hiscox platform along with a package of data for final assessment. If the transaction is completed the insurance is set up and the details (along with documentation) are stored automatically within FreeAgent.

Looking at PI, employer liability, and public liability in the first phases, this level of integration involves overcoming considerable regulatory hurdles as well as technological ones. 

“From a public API perspective, this goes way beyond what most integrations involve, but given these are highly regulated areas, the customer experience has to be absolutely correct,” explains FreeAgent CCO Kevin McCallum. 

For example, FreeAgent is assigned as a formal Appointed Representative to Hiscox, complying with various regulations and procedures, in order to be able to achieve the level of integration required.

As the system develops, McCallum anticipates the increasing level of sophistication. It will become “increasingly smarter, with the ‘intelligence’ to be alerted to new employees joining or revenue increases that might impact policies”.

While some way off, the recent Covid experience can show the impact of lockdown on driving insurance and how this kind of approach might act to safeguard or save costs to a business.

Systems that do the ‘thinking’

Systems that don’t just pass data, but can inform another are at the heart of the recently launched GoProposal and Xavier Analytics integration. Quality and volume scores from Xavier are used to directly guide pricing. 

Xavier interrogation and analysis of the customer’s Xero data is modelled further and directly corresponds to how an accountant could construct their proposal. For example, Xavier creates accurate averages of transaction types over defined periods and has this served dynamically as a pricing option.

As a practising accountant, Xavier founder Jonathan Gaunt is keen to accentuate how these deep integrations are designed to collaborate with the accountant. “Everything is seamless, and works automatically – but not necessarily without the input of the adviser”. 

The final step of deciding which pricing metric to use is down to the accountant. “Building a level of friction into the process provides the opportunity for professional judgement,” added Gaunt.

Integration system cooperation

Gaunt sees the opportunities for how Xavier’s deep integration could be expanded to include cooperation with other systems. 

“I think the GoProposal data is the very start of the customer journey, and where traditional practice management software starts with the need to do a VAT return, we now know a more precise quantum of the job.” 

The result would be an improved dialogue with practice management software to look at the resources, time and staff that would be needed. The ability to project and have multiple levels of communication between systems means that quality, pricing and capacity planning are all part of the same integrated process.

A vision of the future

Where this might lead us is intriguing to think about. Businesses like Codat that specialise in connections between fintech systems see a connected future between systems at the heart of customer experience. 

Codat head of product Yasamin Karimi also sees this as a simplification.“Yes, you can do lots of integrations, but what brings it to life is to pull them all together and [have] apps seamlessly work together. Once the connection is made, and consent has been given, the sharing of data means my interaction with all these bits of technology should reduce”.

The vision of how deeper integrations could work might be more comforting to those unconvinced that true AI will arrive anytime soon or be deployed effectively. The mechanics of the “black box” of how these systems will work still seem tangible and have the ability to build in accountability and even ethical checks, which is something that still raises debate

Given the level of cooperation between businesses to make these kinds of integrations work, over the more ‘plug and play with’ public API approach, shows that when it comes to business the deeper the relationship the better the end result.

Replies (0)

Please login or register to join the discussion.

There are currently no replies, be the first to post a reply.