Data Action Layer June 15, 2020
451 Research found that 55% of financial firms use AI primarily for process automation—to good effect: over 30% of AI adopters cite the benefit of automation to improve productivity; more than one third cite cost savings. Said Jay Leopold of Ameriprise:
The first mile that most business data has to cross—over 75% of it, by some estimates—is the email inbox. Trying to efficiently extract unstructured data from emails and attachments to provide customer service and digital assistance is a critical component of automating business processes.
Raja Doddala, Managing Director at TIAA, concurred, saying that TIAA portfolio managers’ workflows include reviewing thousands of prospectuses.
If a machine could read them and surface actionable insights for a portfolio manager, that portfolio manager would be 100 times more effective. It’s as if you have access to a thousand interns.
Another useful process automation is unifying information from disparate, legacy systems—including calendars, telephone calls, and even documents in the mailroom. Recounts Doddala:
It’s low in the cognitive value chain and it doesn’t sound very sexy, but it’s critical. Even a basic integration and automation of middle office and back office functions can contribute gains.
Any task that can be done by a computer in a data center shouldn’t even have an interface. Just have it done. That is the task of the day.
AI receives a lot of buzz, but this star treatment means there’s an unrealistic expectation from AI, according to Satish Mandalika, Managing Director at AQR. In reality, AI doesn’t have to be exotic to create value.
Instead, opportunities abound to automate simple, manual tasks that clutter daily workflows and perpetuate business processes that have been built in the pre-digital era. Most work still gets done in email, where employees spend too much time on low cognition and repetitive data processing tasks such as inbound contracts, RFPs, fund fact sheets, etc. Recounted Jay:
Coming to the real-world problems of operations, it’s very simple… intelligent behavior is more about simple situational understanding than complex reasoning. It’s really important to have contextual awareness of the problem you’re trying to solve.… Intelligent solutions are more about simple, situational problem-solving as opposed to implementing very complex logic.
For all the potential value AI can unlock, it won’t amount to anything if it isn’t embraced by the people it’s designed to help. Satish added:
How do you enable effective human-and-AI interaction, and when do you have a human feedback loop so the system constantly learns? What is the automation, what is the business workflow, and at which point in the workflow do AI tools come in and help? AI tools need to be embedded into the human workflow.
AI’s most valuable contribution to changing how work gets done in the enterprise has less to do with the AI technology and much more to do with how employees experience it in their workflow. An effective AI solution automates away the inefficient parts of work, simply, in harmony with the employees, delivering transformational change.
Said Alkymi CEO, Harald Collet:
Take out the repetitive work and leave the cognitive work to the subject-matter expert. That’s the partnership between humans and AI that is the recipe for freeing up the next level of productive automation. When the data is important, you want to gracefully involve the employees to assist AI on harder tasks. We see Alkymi as the co-pilot for the subject-matter expert.
To learn how Alkymi can make your workflows flow better, request a demo.
Fine-tuning is not the only way to get relevant, domain-specific responses out of an LLM. Alkymi’s team of expert data scientists explain an alternate route.
Find out which type of automated document processing solution is right for you: data extraction, an IDP, or a complete business system for unstructured data.
We’re partnering with Portfolio BI, a provider of portfolio analytics and reporting solutions, to bring structured and unstructured data sources together.