8 Ways AI Will Make Your Workflows More Productive

If you aren’t using – or at least thinking about using – artificial intelligence to improve your workflows, start. The business reasons for adding AI to workflows are the same benefits we’ve been touting about workflow automation for decades: improved accuracy, better customer service, reduced manual work increased efficiency, better collaboration, scalability – you know the list.

Of course, we’ve all heard the old saying about how automating a poor workflow will just screw things up faster. Now insert artificial intelligence into the equation. The potential for things to go sideways is vast. However, as long as you understand AI limitations and take the time to think through use cases and implementation, AI can supplement workflows, removing repetitive tasks people hate so they are now free from drudgery to focus on “thinking work.” 

This quote from McKinsey, written five years ago before everyone fully boarded the AI hype train, is an excellent summary of what the AI/workflow combination will bring:

Machines will be able to carry out more of the tasks done by humans, complement the work that humans do, and even perform some tasks that go beyond what humans can do. As a result, some occupations will decline, others will grow, and many more will change. 

Ready for your humans to have an AI partner in their workflows? Read on. 

A quick note on AI

If you haven’t yet, remove visions of iRobot, HAL 9000, or C-3PO from your perception of AI. We are still in the first level, narrow/weak AI phase of the technology’s development. These are narrow intelligence tools that can carry out one task – machine learning and natural language processing are examples. General intelligence (strong AI) is the next step up and is human-like intelligence. The groundwork is being laid with supercomputers and generative AI. Artificial superintelligence only exists in science fiction.

There are also four functional types of AI:

1. Reactive machines – Your “we think you’ll like” recommendations.

2. Limited memory (we are here) – These are AI systems trained on an existing body of knowledge (capture professionals have been doing this for decades with large language models, LLMs). This provides a reference model to solve a problem, like a chatbot helping a customer. Almost all AI talked about for business is based on this technology. 

3. Theory of the mind – The AI tool can pick up on emotional queues. Given how bad humans are at this, the machines aren’t even close to this level yet. The closest is sentiment analysis, which is based on using LLM training to identify trigger words.

4. Self-aware – Not even close.

Workflow and AI – how it’s being used now

Even while AI is “only” a bunch of 1s and 0s being rammed together in massive combinations at massive speed, there are some rock-solid use cases emerging. Here are examples and illustrations of where AI is being used to make workflow automation even more effective. 

Accounting workflows

AP and invoice automation remain low-hanging fruit for automation. Workflow plus AI can streamline many aspects of an organization’s financial processes:

• Automate invoices. In use for decades, optical character recognition (OCR) and natural language processing (NLP) extracts information from an invoice, classifies it, and inputs the data into the correct system or database. 

• Manage expenses. Analyze receipts, categorize expenses, and flag out-of-policy expenses. Combined with predictive analytics, this can forecast expenses to help with budgeting.

• Accounts payable and receivable. Automating payments and matching invoices to purchase orders improves efficiency. Also analyze data and payment terms to optimize timing for payment so you aren’t paying too soon but are also able to take advantage of all early payment bonuses.

• Tax prep. Speed tax filing by classifying and analyzing transactions to automatically apply relevant tax codes and calculate tax obligations. Stay up to date on tax law changes for real-time compliance too. 

• Automated reconciliation. Match data across documents and accounts to flag inconsistencies for human review, dramatically increasing the speed of tedious, time-consuming manual processes.

Intelligent document processing vendors are all actively adding AI to their products, often through partnerships. Advances in machine learning over the last handful of years have vastly expanded the type and volume of documents that can be recognized and automated, which has historically been a huge limitation for completely automating all invoices. 

Predictive analysis

Knowing what to do next based on AI-assisted analysis of an organization’s data has huge potential benefits. AI supports predictive analysis in a few different ways:

• Data collection and processing. An AI tool can aggregate gigantic amounts of historical and real-time data from within workflows: transaction records, customer behavior data, operational statistics, market trends, etc. Once collected, this data can be analyzed for patterns and relationships. 

• Model building/training. Machine learning can be used to “teach” a model potential cause/effect patterns based on historical data. 

• Continuous learning. As the AI system is fed more data, predictions improve over time.

• Automated decisions. Some actions can be automated, such as adjusting pricing, reordering stock, or sending alerts.

In finance, predictions could include market trends or credit risks; in healthcare, identifying potential disease outbreaks or patient health outcomes; or in retail, predicting a change to consumer buying patterns. 

Sales and marketing

AI is being rapidly adopted in sales and marketing to support a variety of workflows. Companies have been using AI for lead scoring and personalized recommendations for years. 

Chatbots are early examples of AI in use. They are commonly used to answer simple questions and automation, guiding a potential customer through the sales funnel with recommendations, answering questions, or helping with service issues. The benefits are twofold: the customer gets answers faster, and employees have more time to do more complex work – and answer customer questions that the chatbot can’t. 

Social media monitoring and sentiment analysis is another opportunity for AI. Companies have been using ML for sentiment analysis for at least the last 10 years – using keywords to get a handle on customer satisfaction. As the tools improve, sentiment analysis can be combined with chatbot workflows to elevate customer replies. A chat that starts with “I’m not happy” could be immediately moved to a customer service rep to respond to while a more general question will run through the regular chatbot workflow.

Generative AI

Many process mining and workflow vendors are using LLMs to allow a conversational approach to business insights. For example, process-mining company Celonis demonstrated a use case of asking questions about process data to generate useful information by allowing non-technical users to create queries with Celonis’ Process Query Language. 

Generative AI is roiling the marketing industry as professionals wrestle with how to use tools to speed up the writing process – blogs, sales emails, customer chats, books, video scripts, literally anything with words (which is everything because creating images with AI tools is a piece of this too). After a year, smart marketing teams are beginning to use generative AI tools for first drafts and ideation. 

Organizations are also using machine learning to create large language models of their own content or of industry-specific content to further improve results. For instance, Google has created Med-PaLM 2 specifically to help answer medical questions and arrive at a diagnosis faster. This could be used in remote locations by medical professionals to query the model to diagnose a health issue when the symptoms presented are unfamiliar.

Proactive maintenance

Especially in conjunction with process mining, AI can be used to identify potential equipment failures before they happen. Rather than waiting for something to break, manufacturers and IT departments can avoid user and/or customer aggravation by maintaining equipment and systems based on usage data. 

Workflow discovery/analysis

Another process mining-focused use case, AI can improve discovery via algorithms to interpret huge data sets to identify patterns, ad-hoc processes and other deviations, and bottlenecks. When combined with advanced AI models and machine learning, continuous improvement is possible as data is used to continually improve process map accuracy.

Compliance/risk management

Continually monitoring processes allows a company to adhere to regulatory guidelines and requirements. Uncover potential risks by flagging possible compliance breaches. 

Managing tasks

The basics aren’t sexy but deliver value. No surprise here, AI can automate repetitive tasks – meeting scheduling, using RPA to move data between systems, and routing documents for approval, are all things workflow already improves. 

What’s next

If you haven’t begun to consider using AI to assist in your workflows, it’s time to start. A few tips:

1. Ignore the bright and shiny stuff. Focus on what you can practically accomplish with the technology available today. Don’t get distracted by gauzy promises of the future in favor of what you can do today. 

2. Talk to your workflow/process automation vendor about their plan to include AI tools that goes deeper than a press release.

3. Begin to identify business cases within your organization.

4. Remember that AI is just another IT tool. Like every other hot, sexy new IT tool, it will not solve your business problems all by itself. 

This article barely touches on the potential of AI and workflow together. The reality is that there are so many unique nooks and crannies in workflows depending on the unique needs of an organization, its history and culture, and applications that it’s impossible to identify all the ways AI can boost workflows. 

My best advice: be alert to possibilities. Even if the technology isn’t quite ready yet to implement your workflow idea, it might be tomorrow. Talk to your workflow software provider and understand how they are going to implement AI within their tool – and how that impacts your business. 

One final thought: even if you haven’t started, there’s plenty of time to catch up. Just don’t delay too long – your competition won’t.