Evolve Past “Workslop” With Practical, Human-Centered AI

The “AI slop” problem has generated a fair amount of cultural buzz and media attention over the past few years as the use of LLMs and other AI content generators continues to skyrocket. People notice when low-quality images and substandard prose flood their social feeds.

Thanks to AI slop, we’re now less likely to trust ad content that we suspect is AI-generated, even if it’s not, and readers are picking up tell-tale signs of LLM-generated content, such as the overuse of em dashes. Unfortunately, “workslop” is now a thing too.

What Is Workslop, and Why Should Finance Leaders Care About It?

Every CFO knows the frustration of chasing down a budget variance or spending hours reconciling unexplained anomalies. In today’s enterprise landscape, the promise of AI is everywhere, but so is a new productivity killer: workslop.

Workslop is the automation byproduct that looks polished but lacks substance, context or utility. It’s the article littered with em dashes that doesn’t teach you anything new; the generic report that raises more questions than it answers; the approval workflow that creates friction instead of clarity. It’s the AI-generated content that forces finance teams to do more work, not less.

Workslop is most often associated with poor content quality. It devalues the brand, is less trustworthy and sends the message that people have stopped paying attention. But when workslop starts affecting business applications like the ERP, it becomes even more of a drain on productivity and trust.

Workslop results when AI systems generate output without enough human input, context or oversight. For finance leaders, this means spending valuable time clarifying, correcting or reworking what should have been automated.

The result? Lost efficiency, diminished confidence in automation and a finance function that’s stuck in reactive mode. You may think your organization isn’t invested enough in AI to be affected by workslop, but it’s out there already.

A recent HuffPost article cited a Stanford University study that found more than half of workers say they’ve encountered workslop on the job. In addition to annoying the affected workers, workslop threatens to undermine the key selling point for integrating AI into the workplace: greater productivity with exceptional quality.

The good news is you can minimize or even eliminate workslop with a practical, human-centered approach to AI. Here’s a look at the current state of the workslop problem, what a more thoughtful application of AI technology in the workplace can look like and some tips on achieving an agile, iterative AI deployment.

What if Workslop Isn’t a Problem but Rather a First Draft?

Let’s face it — it’s nearly 2026, and AI is an exciting product. It has incredible potential to save time and improve productivity, so people are going to use it, whether their employer encourages them to adopt the technology or not. The question is, will they apply it with the appropriate training and effort necessary to get the best results?

Workslop happens when the user doesn’t give AI sufficient or well-structured input. To get the best results with AI, you have to keep the conversation going. You have to rewrite your prompt or refine your needs. This back-and-forth process introduces more context and feedback and helps you land at a better result.

I found this out firsthand when I created an AI prompt I envisioned as an end-of-day ritual to update my to-do list by summarizing unanswered emails and flagging commitments I’d made. It sounded like a great idea, but the original version was too overblown and heavyweight to be of any practical use at all.

It took a lot of refinement, feedback and coaching from the LLM to get to a predictable and practical output. It required me to be clear about my needs, information-processing style and attention span to get a result that worked.

It would be fair to call my first draft “workslop,” but through refinement, I got to a useful AI tool. But what if I had stopped at the first iteration and stuck with the less user-friendly first draft? Had I done that, I’d be dealing with workslop that hampered productivity.

Magnify that out across more complex processes involving multiple parties, and you can easily see how AI applied with the best intentions can become workslop — unless you have the training, perseverance and grounding to make it effective.

There is no doubt that AI can add real value. But as leaders, we need to make sure employees have the know-how, support and coordination to succeed, and reports from workplaces on the frontlines indicate there’s still a lot of work to do.

What’s a Human-Centered Approach to AI, and How Do You Get There?

So, what is a human-centered approach to AI? And how can a practical path lead to better results as AI is integrated into workflows?

For AI advocates in the workplace, a good starting point is to acknowledge that the goal isn’t to replace people. It’s to ease friction and amplify our intelligence by understanding the human: their needs, their daily annoyances, their judgment and their goals.

There are two lessons here to bring human-centered, quality AI to the workplace. First, for your teams working with generative AI, make sure they have the training and time to get better results with strong context and refinement.

For the systems you choose that offer AI enablement, make sure your technology partners really understand your team’s needs. That means understanding their day-to-day operating environment, what works and what still frustrates them.

What Does Human-Centered AI Look Like in the Workplace?

AI can be applied on a standalone basis to make people’s jobs easier or used to augment older technologies that leave irritating gaps in workflows. Take Optical Character Recognition (OCR) technology, for example. It converts images of text into readable, searchable text and has been used for years to streamline tasks like entering paper receipts or invoices into expense reporting software.

But as anyone who uses OCR regularly knows, it doesn’t always work as advertised. Maybe you took that photo of a receipt on a moving train, and the receipt was bent, obscuring information. Maybe the invoice is written in someone’s illegible handwriting. Perhaps the date is in European format, and the system only recognizes U.S. format.

There are countless reasons why OCR can fail to translate data correctly. It’s a limited technology. Integrating a more sophisticated technology like AI can close those gaps and finally eliminate the annoyance of having to manually enter those figures.

That’s just the beginning of what human-centered AI can make possible. Given AI’s capabilities, new applications can do so much more to alleviate friction on the job. For example, with the right prompts and thoughtful historical transaction data pattern recognition, AI will be able to add context to an invoice beyond the fields on the page by inferring cost center, project information and more via context centered on the human who is using it.

Human-centered AI can also alleviate workplace friction by taking tasks to people outside of systems like the company ERP. Most people’s jobs don’t live in the ERP system, but they have to log into it (and other systems) to do specific tasks like approving time sheets or employee requests.

What if an AI agent brought those tasks to the person instead, along with the relevant context they need, to make a decision in a program they are already using? That could keep processes moving and employees more focused. Human-centered AI of this type can eliminate non-value-add tasks like data entry and logging into multiple systems.

How Is Human-Centered AI Transforming Finance Functions?

An agile, iterative approach to AI is already transforming finance functions in significant ways. When finance professionals are knee-deep in spreadsheets and analytics, it can be difficult to shift the story-telling side of the brain; so why not build an AI agent to aid in providing that context?

For example, aberrations and anomalies are a chronic irritant for finance professionals, and AI can pick up the slack by providing context to explain spikes in corporate spending. A well-designed agent can flag potential problems before the finance analyst digs through all the spreadsheets to discover the variances.

Similarly, agile, iterative AI can flag anomalies before they bubble up in the HR space. When there’s a variance in pay after a payroll run and an employee questions it, someone on the HR team must drop everything and perform a forensic analysis to discover the reason for the difference. That’s a real challenge for busy teams.

A thoughtfully designed AI agent could surface aberrations before employees are affected, flagging the anomaly and providing context to HR decision-makers where they are. In this way, team members’ focus remains on maximizing productivity rather than putting out fires, and operations run more smoothly.

Eliminating Friction and Workslop: DIY Agents or Vendor AI?

The best way to avoid workslop and get real value from AI is to look for ways to reduce the daily dose of annoyances we all encounter in our jobs by taking on tasks that don’t add value. For some employees, including many finance and HR roles, typing data into a system is an annoyance that can often be eliminated via thoughtful automation.

For people who create content, typing is part of the job, but leveraging AI effectively requires training, collaboration and policies that help employees craft prompts that generate meaningful content and don’t create downstream work for colleagues.

For work automation, the right solution will vary by role and industry, but leaders who are integrating AI into the workplace will often have to decide whether to create agents themselves or obtain an out-of-the-box AI solution from a vendor.

For companies with robust IT resources, including unfettered access to AI expertise or a system integrator on retainer, the sky is the limit. In that case, a vendor that delivers agent-building technology clients use to create AI solutions directly might work.

But many businesses don’t have access to those resources, and even if they do, workslop can quickly become a problem when people try to build their own AI agents without the proper training and resources to avoid pitfalls downstream.

Security is another critical consideration. Keep in mind that people are going to use AI, period. That means it’s the leader’s task to make sure employees are using it safely and transparently — and without introducing chaos.

What Should You Consider When Selecting Vendors?

For many companies, an AI-enabled system from a vendor is a great option, but remember that not all products are created equally. The best way to avoid workslop and get real value from AI is to find a system that knows you as intimately as possible.

For example, if your goal is to improve operations with an AI-enabled ERP system, consider these questions for prospective vendors:

• Does the product eliminate the friction your employees encounter most often?
• Does it solve the most difficult problems your employees face?
• Can it accommodate varying levels of expertise within your organization?
• Does it keep humans in the loop and ensure accountability and transparency?

Whether you’re using a system to generate content, automate workflows or answer questions, the quality of your results depends on how much the system knows about your context. Ask your technology partners how their AI solutions center the human and deliver real value.

Is Workslop Inevitable?

Regardless of who your vendor is and whether you are building your own agents or using a solution that removes friction via out-of-the-box automation, it’s up to you as a leader to make sure AI is secure, transparent and adds value.

Keep in mind that human-centered AI isn’t solely defined by whether it solves real problems and makes people’s jobs easier. Practical, human-centered AI also keeps humans in the loop because, ultimately, we humans are accountable for the results.

Workslop may be an inevitable stage of AI evolution, but it doesn’t have to be a permanent fixture in your finance function. By centering humans in the loop, investing in training and choosing vendors who understand your business context, CFOs can unlock new levels of productivity and strategic value from ERP systems.

The next wave of ERP innovation will be driven by AI that understands your business as well as you do and is capable of delivering insights, automating routine tasks and empowering finance leaders to focus on what matters most.

The future of finance is context-rich, agile and human-powered. You deserve tools you can use today to take you into tomorrow, and you can evolve past workslop with practical, human-centered AI to reach that destination.