How To Prepare Insurance Data For Generative AI Prime Time

How To Prepare Insurance Data For Generative AI Prime Time

By Robert Clark

If someone had a Formula One race car that was not reaching its full performance potential, they likely would check the powerful engine and look into the car’s aerodynamics—the effect of air on an object and its analysis—which teams spend 20% of their budgets on.

You may be wondering what any of this has to do with insurance data. Like most industries today, insurers are scrambling to figure out how to use generative AI and other advanced AI technologies. This new era of supercharged AI is similar to a Formula One car because without the proper data (similar to aerodynamics), the driver will be unable to make the car reach its optimal performance no matter how strong the engines are.

This problem is what many insurers who have taken the leap with AI have discovered in their excitement to jump in the driver’s seat. While there is no doubt insurers should be getting ready to race, they should thoroughly check the status of their data to ensure they’re ready to start the AI supercar.

In this article, I will walk through the fundamentals that insurers need to think about if they want to deliver next-generation customer experiences and products that can transform their growth potential.

Five Short Steps

Getting insurance data in better order is the most crucial step for insurers looking at ChatGPT and other solutions. As I covered in my last article, as ChatGPT responded to my inquiry, it can only be valuable for insurers once a customized dataset is plugged into it. To adjust a common phrase, ChatGPT is what it eats. (I think this applies to all industries, not just insurance.)

1. Start identifying relevant data sources.

Insurers must identify and collect relevant data across all their core systems. This is a quick task for smaller carriers, but for a large insurer, this requires immense internal resources across claims, underwriting and other departments.

This step also should include pertinent third-party external data that rounds out the data landscape. While it should go without saying, please ensure online and offline data sources and systems are secure.

2. Deep clean your data.

This step is critical to what ChatGPT or other AI systems will output. Insurers need to identify irrelevant data and find and correct errors. After that, they can then develop a standard format that will be fed into the AI system. Cleaning and preprocessing data are like preparing a high-quality meal for a premier athlete, but in this instance, the AI is consuming the feast of data.

3. Categorize your information.

Now that an insurer has clean and standardized data, they need to segment the information into categories about their various business functions (claims, customer data, etc.). For example, “this labeling will enable insurers to better analyze claims for type, cost and location to predict the impact on their revenue reserves” or “this segmented data will plug into an external data partner that will integrate via API into a Generative AI tool like ChatGPT.”

4. Have good AI values.

Assuming an insurer did the previous steps correctly, one of the most important things to do next is to establish standards for ethically using customer data. The last decade, even until this past December, was fraught with examples of bias in AI. With many socioeconomic factors making today’s world a very delicate but sophisticated place to live, companies have an opportunity to use AI for good. That good can be customer satisfaction, revenue growth or social impact for non-profit ventures. Whatever the activity, the people using the tools must have good AI values.

Insurers have led this charge before the tech boom of the 2010s by being responsible with customer information and relationships. There is no better time for insurers to step up and not sacrifice ethics to be the insurer with the fanciest or smartest AI solution that then turns out to be a source of discrimination.

5. Get your feet wet and keep going deeper into the water.

Some insurers already know what AI platform they want to use, generative or not. However, for those that do not, they should conduct thoughtful research that considers what experts are saying in mediums like this article.

Additionally, they should not assume that only the new AI tools from the most prominent names are the only options out there. They should look at the relevance to their industry from the company that built the AI tool. For example, an insurance-specific tool will likely have a leg up on a generic supercharged AI tool in delivering long-term customer value because it will not need to be constantly trained to put the proper insurance context on AI results.

After an insurer has made their choice, then it is about training, testing, monitoring and optimizing models while keeping their good AI values in mind.

In Conclusion

Based on my experiences working both on the carrier and technology vendor sides, this approach to data preparedness is a difficult task that many insurers need to invest more time in. By prioritizing data (the parallel to our metaphorical aerodynamics), insurers can best be prepared to handle the Formula One AI supercar. In turn, this can help to improve internal operational efficiency, boost profitability, identify new product or market opportunities, transform the customer experience, and ensure scalable business growth for years to come.

Read the full article.

How the generative AI boom could deliver a wave of successful businesses

How the generative AI boom could deliver a wave of successful businesses

By Christian Owens

Generative AI (Gen AI) is the buzzword of the year, gripping the global tech ecosystem. Leading VC Sequoia declared that gen AI could “generate trillions of dollars of economic value,” and thousands of businesses, from Microsoft to Fiat, have raced to integrate the technology as a way to speed up productivity and deliver more value for customers.

Any nascent sector like generative AI, as was the case with Web3, also brings with it plenty of predictions about just how big it can/will become. The global AI market is currently worth $136.6 billion, with some estimating that it will grow by 40% over the next eight years. Even an overall slowdown in VC dealmaking has made an exception for Gen AI, with AI-assisted startups making up over half of VC investments in the last year.

However, although generative AI tools are attracting headlines and frugal VCs’ money, and while some of the first movers have developed nifty AI tools that respond to critical pain points, how many of these will go on to become long-term businesses? Most that have monetized have stumbled into becoming businesses rather than as part of any long-term strategy, so what will they do if/when they need to scale to meet demand?

There’s a lot that Gen AI startups still have to do to take this captivating technology and actually turn it into a sustainable business. In this article, I’ll explain where generative AI startups can start if they want to turn this short-term hype into long-term growth so they don’t miss a potentially huge market opportunity.

First, it’s difficult to take a new technology and actually turn it into something profitable. While Gen AI tech is certainly impressive, it’s unclear how to monetize or integrate it into a profitable business model. So far, some of the most successful AI startups have used the tech to boost operational efficiency — like, which automates repeating processes that drive revenue and retention — or to help with language processing and content creation, like AI copywriting assistant But you can only have so many AI chatbots. Emerging Gen AI startups will have to carve out their own niches if they want to be successful.

AI companies will also find it hard to maintain a competitive edge. Many AI startups are already struggling to differentiate themselves in an incredibly crowded market, and for every one entrepreneur with an innovative use case, there are ten more riding the wave with no destination in mind — presenting a “solution” without a clear idea of the problem it seeks to solve. There are already 130 Gen AI startups in Europe alone, and the chances of all of these companies reaching long-term profitability are slim.

Finally, AI is still a nascent technology with big questions about ethics, misinformation and national security concerns to be answered. AI companies looking to streamline workflows will have to address concerns about third-party software accessing potentially sensitive internal data before they can be widely adopted, while startups leveraging the speed and efficiency of Gen AI must come up with sufficient guardrails to address the dystopian concerns that these “machines” could come to replace up to a quarter of our jobs.

Read the full article.

How generative artificial intelligence can make engineers more efficient

How generative artificial intelligence can make engineers more efficient

Future Of Work: How Generative AI Is Redefining The Way We Work

Future Of Work: How Generative AI Is Redefining The Way We Work

How generative AI’s impact on digital advertising methodology is evolving

How generative AI’s impact on digital advertising methodology is evolving

By Ken Harlan

The tidal wave of new generative AI tools is causing industries to reassess how they function and identify ways of up-leveling their processes. The current iteration of AI tools offers users unprecedented speed at creating text and visual assets — obviously an interesting proposition for brands and advertisers. But in the near term, the tools’ real benefits are less associated with brand-visibility efforts, and more on paving the way for innovative solutions and quick campaign ideations.

However, today’s generative AI comes with a trove of potential issues around content “ownership” and brand safety. While the digital marketing industry is poised to adopt the technology, it’s important to consider the most impactful ways generative AI can move our industry forward in the near term.

Realities for ad creative today

One thing brands and advertisers need to consider is the potential for generative AI-created content to closely resemble existing artwork. Because content can be generated and implemented into campaigns so quickly, it’s become very easy for brands and advertisers to unknowingly use imagery and messaging that infringes on intellectual property or copyrighted assets. We’ve also found that generative AI often suggests terms, mottos and slogans that are copyrighted unless asked specifically to remove any copyrighted text.

Another consideration is around brand safety; there’s a risk of generative AI creating assets that do not fit brand guidelines or are offensive to certain audiences. This obviously has brand reputation implications. That said, advertisers need to constantly ensure AI-generated content aligns with their brand values and will resonate with target audiences.

Despite these hurdles, the generative AI market is forecast to reach $188.62 billion by 2032, up from $8.65 billion in 2022. From where we sit, this makes sense. We’re all seeing the surge of interest in AI, and quickly realizing how the current tools represent an amazing “jumping off point” for advancing workflows.

Platforms like Midjourney allow users to develop images simply by typing in basic text. The initial assets it creates, based on your prompt, could turn out to be very close to an image you are thinking of, or could be nothing like you imagined — in a good way. It enables teams to essentially have a very fast, and interesting, brainstorming partner. It opens the door to accidental creativity and inspires fresh perspectives on what branded collateral can be for a campaign.

From there, it’s up to the creative team to carry those assets across the finish line in a way that meets all brand guidelines.

Still a ways to go for code development

Similarly, we’re starting to see generative AI used in developing first-draft code for new digital advertising products or solution updates. When it comes to developing new solutions or evolving existing ones, it can take a few weeks to several months to write and test code. Solutions like ChatGPT deliver first drafts in seconds.

While the speed is very impressive, it’s important to review it for a few critical reasons.

We’ve found that generative AI produces code that is often not optimized for performance or security. Additionally, the code might not be scalable. These issues result in products that miss the mark in regards to reliability standards.

It’s also difficult to maintain, modify and incorporate the code into existing products — and that’s the most impactful drawback at this point. If every digital solution was initially developed by AI, things would likely function properly, and could be easily innovated and updated. But humans developed the initial code, and there is too much variability in how we build solutions. It’s that variability that makes current AI-generated code unable to seamlessly integrate with what we’ve previously made. So, just as with using AI tools for plug-and-play creative assets, we still need a fact-checker or goalkeeper.

Nonetheless, these tools are absolutely here to stay. The quicker we learn their use cases and hindrances, the faster we can optimize our workflows for the better. Only by adopting generative AI tools can brands, advertisers and solution providers understand what’s coming in the new frontier.

Read the full article. 

Generative AI Reshaping Countless Everyday Consumer Touchpoints

Generative AI Reshaping Countless Everyday Consumer Touchpoints

Generative artificial intelligence (AI) isn’t coming. It’s already here.

At this very moment, the technology’s capabilities are already enacting sweeping changes across the ways in which end-users interact with and what they expect from the digital tools that make up much of modern life.

That’s because the latest generation of AI is able to undertake non-closed loop, iterative tasks in real time — a massive upgrade from earlier generations of predictive AI and accompanying machine learning (ML) automations.

Amazon is working to embed the tech’s capabilities within its search function and inventory system, while Google and Microsoft battle it out over who can offer their millions of users the most up-to-date experience across both enterprise and individual product suites.  

Elsewhere, startups and mature tech firms alike are increasingly tacking the tech onto their offerings to boost their valuations and streamline efficiencies across operational and back-end touchpoints.

Even fast-food restaurants are leaning in to AI’s moment, with Wendy’s integrating conversational AI into its drive-thru offering.

As generative AI capabilities are integrated into the broader marketplace, observers predict a dramatic change as digital interactions undergo a transformational shift.

In a sign of the changing times, OpenAI, the firm behind ChatGPT, Thursday (May 18) launched a ChatGPT app in the U.S.

Leveraging the Power of AI

Generative AI, unlike predictive AI, can “generate” or create new content such as text, speech, images, music, video and code.

But the state of hype around AI will likely dwindle — fast — if the technology cannot be used for long-term, practical applications across industries that demonstrate its value.

Given how many enterprise operations, as well as day-to-day consumer touchpoints, have significant software components, generative AI will impact, at least in some manner, how businesses engage with their customers, and how they compete with each other, particularly in marketplaces where speed to discovery can give a firm an edge.

“There is a lot of opportunity to build new user-facing products, or those that better delight users in an existing experience, using AI,” Emily Glassberg Sands, head of information and data science at Stripe, told PYMNTS talking about her company’s use of AI.

That’s because AI will empower firms to tailor all aspects of digital interaction to how the end-user wants it to flow, driving  personalization by optimizing the when, what, and how, as well as the ease of any digital interaction.

Questions like, “When am I next free for dinner in July with these three people?” or “What is my availability to fly to California next week for a meeting?” will no longer require users to check their own calendars or search for flights, and can instead be answered in real-time by generative AI applications able to recognize a query, surface the possibilities, and rank them to generate the best answer in a millionth of the time it would take a person to do the same.

Andrew Gleiser, chief revenue officer at payments provider Aeropay, told PYMNTS that one use case he sees for generative AI is integrating it into merchant payment portals to surface compliant information to customers around their own best clients as it relates to metrics, including average order value, overall volume and purchase cadence.

Making Tasks More Efficient

The immediate impact of AI is that knowledge-based work is being increasingly augmented by the technology’s ability to decrease the time cost of various operations.

Amias Gerety, partner at QED Investors, told PYMNTS that at the moment, the most interesting areas where AI is being applied lie in fraud and risk management and underwriting — because those are industries where practitioners are sophisticated and can use AI as tools to get fast answers while ensuring themselves that results are accurate.

In his testimony before Congress earlier this week (May 16), OpenAI CEO Sam Altman gave the real-life example of a dyslexic small business owner who created an AI tool to automate the drafting of professional emails that resulted in several hundred thousand dollars worth of new business.

AI is a tool — not a creature. “It will do tasks, not jobs,” Altman told U.S. lawmakers.

He provided additional examples of how the educational nonprofit Khan Academy is piloting a program that uses GPT-4 to power a personalized virtual tutor for students and a classroom assistant for teachers, while Morgan Stanley is using GPT-4 to power an internal-facing chatbot to help financial advisers better serve their clients, and Weave is using generative AI tools to build a collaboration platform for scientists, specifically focused on breakthroughs in oncology, among other use cases.

The most common use of ChatGPT at the moment is through text-based prompts, but PYMNTS has been tracking how the greenfield opportunity may be to marry speech and commerce.

PYMNTS research has found that there are more than 86 million consumers in the U.S. alone using voice assistants.

As PYMNTS has previously reported, while business interests continue to push AI forward, many academics, technologists and researchers continue to warn about the technology’s danger absent appropriate guardrails around its use — including Google’s own Geoffrey Hinton, who resigned in advance of the I/O conference in order to “speak freely” about the potential risks that widespread integration of the technology may pose.

Read the full article.

The 4 categories of AI that impact marketing: Generative AI

The 4 categories of AI that impact marketing: Generative AI

By Greg Kihlstrom

Generative AI For Business Leaders 101

Generative AI For Business Leaders 101

By Amit Ben

As a technology and business leader, I have witnessed numerous technologies emerge over the past two decades. While many of these have had a significant impact on businesses, few have truly transformed our society and culture. Examples of such transformative technologies include the personal computer, the internet and smartphones. It is widely accepted that artificial intelligence (AI) is one of these transformative technologies. In this piece, I aim to help business leaders accelerate the success of their AI journey by offering an overview of the technology and its limitations, as well as providing practical tips to consider when implementing generative AI in products and business operations.

Technology Overview

What is “AI”?

AI stands for “artificial intelligence,” which is the simulation of human intelligence in machines to perform tasks that typically require human intelligence, such as decision making, perception and language translation. AI is powered by various techniques and has a wide range of applications across various industries, including healthcare, finance, transportation and entertainment. As AI continues to advance, it is expected to have a profound impact on society and shape the future of work, education and daily life.

What is generative AI?

Generative AI is a type of artificial intelligence that generates new data based on a set of inputs. Generative AI has applications in creating music, art, literature and other fields. The main challenge is to produce realistic and desirable outputs.

Why has language AI lagged behind other AI domains?

Language AI, or natural language processing (NLP), has matured more slowly than other types of AI, such as computer vision, due to the complexity and ambiguity of language. Words can have multiple meanings, sentences can have complex grammatical structures and language can be highly dependent on cultural and social contexts. However, recent advancements in machine learning and deep learning, along with the availability of large datasets and computing power, have led to significant progress in NLP, particularly with the development of language models such as GPT-3.

What are LLMs, and what are they good for?

LLM stands for “large language model,” pretrained on vast amounts of text data to generate coherent and contextually appropriate text in response to prompts or inputs. Examples: GPT-3, BERT, RoBERTa. LLMs have a wide range of applications, including translation, question answering, and text generation for summaries, emails, social media posts and chatbots.

What are the limitations of these models?

LLMs have limitations that need to be addressed. They can generate inconsistent and inaccurate results due to noisy or biased input data, and outputs may lack structure, requiring post-processing. Scalability is also a significant challenge as LLMs demand substantial computational resources and may be slow. Lastly, explainability remains a critical concern for LLMs. Understanding the reasons behind a specific output can be difficult, posing challenges in interpreting and auditing these models. Therefore, it’s crucial to carefully evaluate their performance and limitations before incorporating them into production systems.

Risks Of Generative AI In Products And Workflows

When adding generative AI capabilities to their products, services and workflows, businesses and product companies face several risks to consider. Biased or inaccurate outputs can damage the company’s reputation and result in legal or regulatory issues. Lack of interpretability and transparency can make it difficult to understand and explain AI-generated outputs. Data privacy and security breaches, as well as intellectual property theft, are also concerns. Over reliance on AI can lead to neglect of human expertise and judgment. Generative AI also has limitations, including the reliability of the output, the need for customization and fine-tuning and high costs. It can be slow and not suitable for large volume deployments, with limited input size and unstructured output that can be difficult for developers to work with.

What Are The Opportunities?

By combining generative AI with other NLP solutions, businesses can create a comprehensive AI system that is tailored and trained specifically for their needs, with faster time-to-market, lower costs and greater scalability. It’s important for businesses to prioritize explainability and understandability of the underlying data when using generative AI, as this can help ensure that they can trace which source data is behind each part of the generated output.

Let’s examine a few examples that demonstrate the versatility and practicality of large language models in various applications. In the realm of customer success, LLMs can be employed to summarize every customer interaction and extract pertinent information. This capability benefits both the operational level, by maintaining CRM hygiene and saving agents time, and the managerial level, by delivering automated analysis of the Voice of the Customer (VoC) for the C-suite. Another example is building a GPT-based assistant that taps into your company’s knowledge base and triggers specific actions within your product, such as helping customers find and implement solutions to their problems through filling out relevant forms, changing configurations, or locating and acquiring suitable loans.

Additionally, integrating smart CRM systems can streamline operations, as a business user can instruct the CRM via a chat interface to set up a specific pipeline. For example, the user can request that each customer from a particular source meeting certain conditions be tagged and routed to a specific agent. Lastly, LLMs can transform user interfaces, such as forms and web apps, into more flexible and intuitive chat forms. A prime example is a traffic ticket payment process where, instead of navigating numerous forms, a user provides their name to a chatbot that locates the ticket and facilitates payment through the same interface.

Read the full article. 

Business leaders investing in generative AI, automation to reinvent physical operations: Report

Business leaders investing in generative AI, automation to reinvent physical operations

By Victor Dey

Connected operations cloud firm Samsara today released a report that highlights how organizations in industries that drive over 40% of the global GDP are revamping their physical operations.

The 2023 State of Connected Operations Report, compiled after surveying more than 1,500 physical operations leaders from nine countries, reveals that these leaders are making substantial investments in digitization to enhance their supply chains, improve employee skills and adopt sustainable practices, all of which have yielded positive results.

According to the report, challenges faced by operations leaders in the past year garnered significant attention and sparked discussions in boardrooms worldwide. These challenges included soaring fuel costs and inflation, shortages of labor and equipment, and constraints in supply chains. However, they successfully navigated these obstacles by embracing technology and finding ways to optimize efficiency.

Automation and generative AI in the pipeline

Leaders are revising their supply chains and technology budgets to address these challenges and build resilience. The study also found that leaders are now embracing generative artificial intelligence (AI) and automation, with 84% planning to use generative AI and 91% automation to modernize their operations by 2024. Additionally, 51% are already using or planning to use autonomous vehicles or equipment this year.

To optimize operations further, leaders are replacing traditional pen-and-paper processes with digital workflows, with 55% of their field employees predicted to depend on digital workflows to carry out their daily tasks by 2025.

“We asked more than 1,500 physical operations leaders across nine countries how they are reinventing their operations. They have over 3.6 million vehicles and assets under management and over 6.3 million employees,” Jeff Hausman, chief product officer at Samsara, told VentureBeat. “Our research found that two in three leaders are increasing their technology budgets this year, indicating they are confident in the ROI of digital transformation. Leaders report benefits like increased net profit and safety as a result of their investments to date.”

To enhance supply chain predictability and efficiency, 59% of leaders have planned to onshore their operations, meaning they will relocate them to their country of origin this year.

According to the study, real-time operations data held a competitive edge and was deemed crucial for decision-making by 90% of leaders.

“Leaders predicted that by 2025, more than half of employees in the field will rely on digital workflows to perform day-to-day tasks. For employees in physical operations, digitizing workflows reduces friction in administrative aspects and adds more flexibility. It’s a significant change from pen-and-paper processes and speaks to the evolution of technology purpose-built for these roles,” Hausman told VentureBeat. “Labor shortages continue to be a major challenge for operations leaders. Consider, in the U.S., the average driver turnover rate is about 90%. At the same time, digitization is shifting the day-to-day employee experience. So we’re at an inflection point: roles are shifting, and the talent pool is tight.”

Independent research firm Lawless Research conducted the 2023 State of Connected Operations survey from February 6 to March 10, 2023. The audience surveyed comprised 1,525 physical operations leaders, including C-suite executives.

Investing in next-gen technologies to optimize efficiency

“Our research found significant changes are underway in the next 18 months. Leaders are investing heavily in digitization to improve supply chains, employee skills and sustainability practices. These investments are all connected to combatting today’s toughest challenges,” Samsara’s Hausman told VentureBeat.

Leaders anticipate that within the next two years, one out of every six employees will be engaged in roles that don’t exist today. To tackle this upcoming shift, over half (52%) of the leaders have prioritized equipping their employees with the skills they will need to navigate emerging technologies.

Notably, the surveyed organizations are projected to invest approximately $7 billion in 2023 to facilitate employee training, upskilling and reskilling initiatives.

“Optimizing the existing workforce and investing in career development is critical to setting an organization up for long-term success. In fact, a key to retention is proving to employees they are essential to the future of the business and providing opportunities for them to build their careers,” explained Hausman. “Our research also found leaders are uncovering new ways to upskill employees with technology. For example, over half will use extended reality and AI to upskill employees in the next two years.”

Samsara’s research also revealed that data plays a fundamental role in every digital transformation strategy, serving as a robust foundation for fostering resilience and gaining a competitive edge. Technology empowers organizations with expedited access to data, and leaders who possess accurate and timely insights are better equipped to anticipate and proactively address potential issues, thereby ensuring seamless operation.

“Almost every leader we surveyed (90%) said having accurate, real-time operation data is critical to their decision-making,” Dana Chery, VP of marketing at Samsara, told VentureBeat. “They’re dedicating substantial resources to ensure they have the technology in place to leverage that data to its fullest, with two-thirds of leaders reporting that they are increasing their technology budgets for 2023.”

Digital transformation for physical operations

Chery added that managing physical operations is complex, and organizations have historically struggled to collect and analyze data to make informed decisions. However, recent technological advancements, such as plug-and-play digital sensors, wireless technology and cloud-based AI data processing, have enabled a significant digital transformation in the past decade.

“From driver safety to back office operations and customer service, it’s difficult to find a role where technology can’t support improved outcomes. It’s a new era for these industries,” she said. “Our research found that physical operations leaders are excited to test technologies like generative AI to see their potential — only 5% said they had no plans to adopt it. This demonstrates the universal need for technology to support the employee experience and increase efficiency across the board.”

The report also highlighted that connected operations leaders, who possess the highest level of digital maturity, demonstrated a six-fold greater likelihood of surpassing their financial goals by 25% or more.

The study found that these leaders are making substantial investments to fortify their organizations and enhance customer experiences. Many anticipate positive transformations and a favorable return on investment within the next 12-18 months.

Additionally, Improving workforce productivity with new technologies is a critical priority for 56% of those surveyed.

“We took a closer look at the differences between organizations that reported the highest level of digital maturity — Connected Operations Leaders — to those at the beginning stages of digitization,” said Chery. “Compared to organizations in the beginning stages of digitization, Connected Operations Leaders are five times more likely to rate the productivity of their workforce as ‘excellent’ and six times more likely to report exceeding their financial goals by 25% or more. The bottom-line benefits of digitization are clear.”

Read the full article. 

Going Beyond Generative AI To Truly Solve Productivity

Going Beyond Generative AI To Truly Solve Productivity

By Nitish Shrivastava

It would not be an exaggeration to say that the advent of OpenAPI (and ChatGPT) has taken the tech world by storm. Since its first demonstration, large-scale companies have rapidly undertaken initiatives to further the integration of conversational AI into offerings. Microsoft notably has taken to integrating OpenAPI into their enterprise suite of products; mainly, the Copilot initiative that brings generative AI to multiple applications like AI-paired programming, word processing and worksheet applications, etc., has been touted as a game changer for office productivity.

When assessing these AI-enabled applications, a couple of things become clear. The collaboration with AI does a good job of introducing point improvements, starting boilerplate, etc., for the applications being used (whether it be a document outline in Word, a skeleton code in GitHub and so on). Introducing the concept of natural language queries frees its users from utilizing knowledge of esoteric scripts and language snippets to generate the required starting information to speed up tasks. As a method of quickly researching available context to integrated applications and using that to generate a summary, today’s applications appear to scale well.

However, on closer inspection, it also becomes clear that there are inherent challenges in applications incorporating AI. The assistance provided is highly contextual and dependent on several factors (one such factor: the efficacy of the training set used to power the underlying models contributes significantly to how accurately this functionality can contribute towards meaningful results). Even in the most optimistic scenarios, though, there are additional issues. Current capabilities may go some way towards fostering productivity, but only in terms of piecemeal recommendations and actions. When balancing productivity and wellness, mere reports and summaries generated by these applications do not do as well as applications that provide guidance based on AI. Applications today do not take a holistic view of the eco-system of applications that are being utilized in a working professional’s life; instead, they are confined to the applications being enabled with AI. As such, their utility is restricted when viewed in the context of balancing workplace productivity and wellness.

The need of the hour is to look beyond generative AI; unless we do this, their usefulness will remain rooted in the applications they are tied to rather than the end user’s needs. At this stage, I propose a blueprint for what a system (that transcends these limitations and truly embraces productivity and wellness) would look like:

Knowledge Web

Imagine a system capable of interacting with any application (work-related, social, or otherwise) in a cohesive manner, providing access to these apps in a consistent interface to the end user. Such a system would allow end users to interact with enterprise applications cohesively to accomplish complex workflows from a single access point. When you factor in the savings in time and effort, the impact on increased productivity (accompanied by increased leisure time for pursuing wellness concerns) becomes apparent.

For example, imagine a customer support organization that requires its employees to deal with multiple points of direct user interactions (emails, Twitter, messages, phone calls), route it to appropriate customer tickets (considering similarities and connectedness of incidents based on history), redirect to diverse resolution teams working on multiple programming and resolution environments, as well as various time shifts and now combine that with business analysts and other vested interests, looking at existing processes and observing outcomes to understand where to direct time, money and resources to improve customer satisfaction. A universal app ecosystem that ingests information from such systems auto-links data points from diverse applications based on pre-configured or evolved learning-based criteria and empowers users to access these analyses and functionality without jumping across multiple underlying systems be a game changer.

Prescriptive Guidance

With generative AI output, it’s important to note that the information provided is usually of a descriptive or informative nature. The end goal is to provide helpful information to users of the system. However, such information should only be considered an initial step leading to the goal of fostering productivity and wellness. The next step must be to empower users and generate prescriptions and guidance. This would have to be tailored not only to the user’s data or the patterns of work but consider other factors like the apps they use, evaluations of the efficacy of specific activities (possibly through a scoring system) and much more to be truly effective.


Finally, it’s important that such applications provide the ability to utilize the powers of generative AI, context gathered from the app ecosystem and guidance systems to generate automated workflows that replace both mundane, repetitive tasks (and complicated multi-step conditional workflows) that such users typically engage in vis a vis these apps. Such a system mustn’t be tied to apps but provide a consistent generalized mechanism for dealing with different scenarios, desired outcomes, levels of intervention at critical steps and more. For organizations, the ability to create workflows applicable to groups of related users, functions or even organization-wide will foster objectives related to the standardization of critical processes, reducing uncertainty and encouraging successful outcomes.

Read the full article.