Moments of Action; applying AI to the tasks that matter

last updated:
Common Tern

Moments of Action; applying AI to the tasks that matter

by:
last updated:

There is no shortage of opportunities for organizations and suppliers to identify those moments and create potentially significant value for each. Managing the data-heavy requirements of the AI to enable them, along with governance structures to ensure their efficacy in operation - not forgetting their accuracy in the outcome - remains a challenge.

Applying technology to assist in the completion of important tasks has always been the primary way in which it’s moved from the theoretical to the functional. The practicality of how that assistance can be applied, has always been key to adopting that functionality. 

As we discussed previously, the point at which interesting becomes useful is when it meets an important task – a “moment of action” – where it can transform the way in which that action can be performed. Whether that is increasing the velocity at which that task can be performed, the accuracy and precision of the output of the task, or enabling the discovery of new ways to approach that task, all have been assigned at some point as suitable for AI to transform.

Moments That Matter

The application of AI to tasks within specific industry verticals creates specific challenges for themselves, even before we look at the tasks themselves. These tasks can be notionally wrapped in vertical (industry), and horizontal (task type) challenges simultaneously. For example, where specific qualifications are central to an employee’s ability to legally perform their job, how might a generic human resources system cope with this task? Can it manage the renewal and automatic filing of the qualification on an annual basis? What is the exception process, and how is it escalated? For many tasks, the vertical and the horizontal are completely entwined. Whilst industry-specific tuning is vital to even approach the completion of some tasks, being able to handle the task in the first place is the initial point of contact for AI. In addition, we also learn why moments is so operative a word to describe the impact of AI on these line-of-business operations; where AI is typically surfaced is at points of decision, where time is valuable. What is the next best action?

Moments that are Visible

Through the integration with horizontal line-of-business (LoB) applications, AI is most visible to employees and service users (e.g., customers, prospects, applicants). It is also customers, prospects, and applicants that will likely be most affected by the decisions that are influenced – if not made – by AI. 

Some of these have become covertly commonplace; a Large Language Model underpins the autocomplete functionality in the Google Doc that I’m currently writing, making suggestions about sentence completion of the structure. It’s important to remember that such models comprise vast quantities of data, the sources outside an organization, and application providers’ control. If that’s not well handled, where it is not clear why a suggestion is being made – or at least how that suggestion was deemed appropriate – the application in which it is being surfaced becomes a mere broadcaster of a message that it did not directly influence. Indeed the lack of real content governance is one of the reasons why large language models themselves are still controversial

More straightforward use cases, such as those around the initial screening of potential hires on behalf of hard-pressed HR organizations, have suffered from similar governance issues. In these cases, it’s not the technology per se that creates problems, but rather the human interactions on which that technology is trained to replicate are beside the point when that is supposed to be the object of the exercise in the first place. This only reinforces that efficacy and governance should be validated a long time before any attempt to scale is approached, especially when they are surfaced so visibly.

Moments that are of Value

It bears repetition that adopting AI technology within LoB operations sees significantly reduced friction, which can most easily be attributed to revenue generation. This is one of the reasons why the organizational LoB, concerned with customer generation and retention, was often the first to see AI applied to their operations. It is also convenient that these operations are also where the most consistently managed organizational datasets are located and data analysts present, which makes the business of an organization’s specific AI model creation likely to be more successful.

Questions such as “which prospects in my sales funnel look to be demonstrating a propensity to buy?” or “which of my customers are demonstrating that they might churn from their subscription?” form part of a set of predictive analytics that helps predict revenue and assist planning. Improved accuracy in revenue reporting will attract the attention of every senior executive, especially those with shareholder reporting duties to attend to; unsurprisingly, interest in these tools remains significant. Their use, however, isn’t always straightforward. 

For very large organizations with customer data at scale ( built up over significant periods), this is potentially achievable with their own data (1st party). For most organizations, however, other data sources are required to provide enough information on which models can be created and tested to perform accurately; most commonly 3rd, party data (but also privately shared 2nd party data sources) is supplied either via AI vendors directly or through explicit data providers. Aside from the governance as mentioned above overhead that external data sources introduce, there is the real potential that the trade in this external data will become significantly restricted in some jurisdictions as privacy concerns become more politically prescient. 

Moments that release Time

Talking about LoBs as amorphous organizational blobs obscure that lobs are just a bunch of individuals, working with noses pressed against laptop screens, trying to do their best at their desks (at home, in an office, or elsewhere) and in the process creating a long trail of interaction data on how they are working. 

Using this in a collective, non-invasive sense to help improve work through AI techniques to unlock efficiency – or perhaps to just make everyday work easier – to release more of the working day is a continued focus, especially around productivity suites. Aside from the previously mentioned “type ahead” assistance with spelling, grammar, and language construction tools, helping workers prioritize tasks (worked you owe, work owing to you by others), assist in the scheduling of collaboration (“you usually invite these people to this sort of meeting”) or help surface all the emails on a given subject regardless of any formal threading is far from cutting edge applications of AI techniques but can be marginal gains that roll-up into something more contributory in releasing time back into a working day. That balance of assistance vs. invasion remains difficult to manage, and while these moments are of a significant frequency in a working day, AI-derived assistants stay for now, at best interesting optional features; frequently demonstrated and often subsequently disabled. 

Picking Your Moment

The need to find horizontal alignment to everyday organizational tasks is far from mutually exclusive from the need to verticalize to meet the needs of specific industries; the more effectively you can meet both, the more rapidly you will find adoption of an AI-enabled application. The moment that matters, in a format that works, that generates an appreciable value that is measurable, is a goal to be repeatedly focused upon.

There is no shortage of opportunities for organizations and suppliers to identify those moments and create potentially significant value for each. Managing the data-heavy requirements of the AI to enable them, along with governance structures to ensure their efficacy in operation – not forgetting their accuracy in the outcome – remains a challenge. As does focussing on whether what is being created generates enough value to prevent it from becoming a novel feature rather than a potential operational pillar.

If you are a buyer or user of Information & Automation management tools and want to understand the application of AI, we have a free – yup, free – confidential advisory service for you. A service that means all you have to do is find a time on our calendar.

If you are a seller of Information & Automation Management technologies and want to explore the use and positioning of AI in your applications, again feel free to reach out and chat with us, we may be able to help by providing some critical but constructive advice and support.

Leave a Comment

We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it.

New Report: IDP Market Analysis 2024-2027