Barclays has run into problems when project managing end-to-end system upgrades, according to Martin King, head of process transformation for risk, finance and technology at the UK bank.
"Where we have had our fingers burned in the past is where we have tried to interface things like bots with many, many strategic systems across an end-to-end process that creates a lot of headaches for technology and setting up testing environments," said King at the Capital Markets Innovation Summit in central London this week.
“We have tried to interface things like bots with many, many strategic systems across an end-to-end process that creates a lot of headaches for technology and setting up testing environments,” he said. “Where actually most of our success in other pilots has been more self-contained where robots are reading data in Excel files and we are pushing data in and out, templates that typically have the same structure - all be it the values of the data attributes themselves - might change on a month to month basis.
When building robotics and workflows, in the first six to nine months a large amount of investment is focused on building the infrastructure of the process in the case that it is standardised, said King, and then the process can begin to scale.
“When we work with vendors, we are working in an agile format where we can deploy stuff in a six to nine month timeframe, very similar with robotics,” said King. “With a lot of that investment we are not generating any capacity which is the grey zone till after nine months. Generally, it takes 12 to 18 months initially to break even.”
In a later panel, Sotiris Manderis, managing director for corporate and institutional digital at HSBC said the bank had been increasing over the past year and a half its natural language processing (NPL) infrastructure.
“We want our people to actually spend more time with clients than on admin tasks,” said Manderis. “The basic business case behind [NLP] was we have our salespeople globally that need to have access to information but they have to go to the system to find the information – they have to know exactly where to go. So now we are using Symphony as a channel, we have our artificial intelligence data science team who put together this bot, with real intelligence behind it that the bot knows where to get the information from.”
Manderis said the bank was increasing its reliance on fintech partnerships.
“We have chosen not to build especially in the areas of technology where you need to be agile, you need to be quick, you need to experiment and bring a meaningful solution to market fairly quickly,” said Manderis.
On blockchain initiatives, Dino Zannetos, chief operating officer for treasury and trade solutions EMEA at Citi said a contained ecosystem is crucial to success.
“That is the challenge with blockchain, unless every bit of the chain or participant in the ecosystem comes along with that technology and goes on the journey the success rate is low,” he said.