How AI Integrated Enterprises are Orchestrating Their Business

Markus Zirn, Chief Strategy Officer, Workato

20 years ago, Amazon emerged. Ever since, the world of brick and mortar retailers has never been the same again. Amazon caused a Digital Transformation wave, one that separated retailers into “winners and losers” depending on their digital readiness.

Now, there’s a new wave forming that will equally create “winners and losers.” Digital Transformation will give way to the Generative AI Transformation. When my teenage son used ChatGPT for his homework, a biology essay, there was no denying that AI can do “more with less.” He just typed in the topic and seconds later, there was an essay that was so good, the teacher would never believe he wrote it. If your competitor in the marketplace uses GenAI and you don’t, you’re clearly at a disadvantage. 

Google was the first to take notice of this new wave and rang the “code red” alarm bell. The BPO (Business Process Outsourcing) industry is next in line to feel the heat as Phil Fersht highlighted. Among all the uncertainty, one thing is crystal clear: the Generative AI Transformation wave will reach every one of us.

Note that AI/ML isn’t anything fundamentally new. Data science teams have long been applying ML (machine learning). However, what’s fundamentally new is that, with GenAI, AI is now accessible to all of us via the English language. AI is being democratized, which is the reason why the AI Transformation wave is being triggered.

Since OpenAI’s ChatGPT became the phenomenon in late 2022, Workato customers have been among the pioneers to leverage GenAI capabilities. This also includes our own “Workato @ Workato” efforts. All of this is made possible via Workato’s OpenAI connector. We’ve seen 500% growth in the use of the OpenAI connector in 2023. Examples of such projects can be found at the “New Automation Mindset” podcast.

Among our customers, we see the following distribution of AI Integrated use cases. Almost half are used for sales and marketing use cases. IT Operations (Helpdesk Chat Bot and IT Knowledge Management) is next with 31%. The remaining 21% are made up of Finance and Support use cases.

Share of Generative AI-enabled Processes Across Business Teams

The business outcomes have been staggering. Customers report exponential improvements with regards to business operations productivity, demand generation, security, customer experience, and more.

At Workato, we describe companies who realize such staggering business outcomes with a combination of AI and Automation as "AI Integrated."

This multiplier effect is why Tom Davenport recently observed in HBR that this new AI era "requires not only an appreciation of AI, but also a renewed appreciation of business processes as a structure for improving work."

Infusing Gen AI capabilities into automations, and therefore into key business processes, makes for a very powerful combination. Furthermore, this combo pushes technology democratization further ahead. Low code orchestration platforms have democratized the implementation of end to end business processes beyond developers coding scripts that call APIs. Equally, Gen AI is democratizing the use of AI beyond data scientist coding machine learning algorithms. The result is smart orchestration accessible to a much larger audience in the company.

So how can GenAI concretely help? We observe the following 3 fundamental “AI Integrated” patterns:

Pattern 1: Operationalizing Unstructured Data

The digitization wave was all about structured data (e.g. online form entries) for business operations. For example, the CRM system keeps records and guides a sales rep through sales steps. The new opportunity with GenAI’s Large Language Models (LLMs) is that now, unstructured data can be operationalized. Any “digital footprint” created (emails, texts, photos taken, voice or video dialogues that were recorded etc.) can be made usable and actionable. To grasp the potential, it’s important to understand that 80% of today’s data is unstructured. LLMs can be trained to understand different languages, so translation can be included in the process of understanding the unstructured data.

This is a great improvement for customer experience as a customer will never have to enter data in structured form that has already been shared in an unstructured dialogue. An LLM can parse any such dialogue, summarize it, extract key topics and also identify trends, patterns and themes. Sales dialogues as well as support conversations are key use cases. 

In addition, an LLM can then operationalize “next steps” based on this understanding. It can trigger follow-up processes (send email, schedule a demo, populate Salesforce with data gathered during the call etc.) and generate the appropriate text for those. 

Pattern 2: AI-Powered Knowledge Management

My first job after college was with Booz Allen & Hamilton, a management consultancy. Back in the 90s, Booz Allen was a first mover with its KOL (knowledge online) system that made all previous project presentations (of course anonymized) searchable. I can tell you that for my own work, this was a game changer.

Combining the power of Large Language Models (LLMs) to generate vector embeddings with the long-term memory of a vector database creates “semantic search”. Unlike traditional keyword-based search, semantic search uses the meaning of the search query. It finds relevant results even if they don't exactly match the query. Hence, semantic search has the potential to enable an entirely new level of knowledge management. 

As we have so much “unstructured knowledge” spread across our enterprises, upgrading from traditional keyword-based search to semantic search has the potential to become the “killer app” for AI in the enterprise. It’s immediately applicable for product support as well as internal helpdesk (IT and HR) and all sort of research tasks. If given enough permission, LLMs can be able to amass the conversed knowledge of entire organizations - putting an end to the time-consuming and costly knowledge silos that are haunting every organization today.

Note that the content doesn’t have to be text. LLMs can also understand photos remarkably well. As LLMs can generate language it’s great at dialogue leading to a new quality level for chatbots. This includes translating between different languages.

Pattern 3: Smart Business Rules Engine

The last era of automation mainly centered around Robotics Process Automation (RPA), which are rule-based scripts instructed by humans. The new era will center around Generative AI in general and LLMs in particular and we will be able to go beyond rule-based scripts to automate more heterogeneous tasks without having to give explicit instructions. So, AI has the potential to unlock a new dimension of automation for intelligent organizations. That’s what the AI Integrated Enterprise is all about.

The third GenAI pattern is about a new, intelligent Business Rules Engine. Older-timers will remember the world of BPM (Business Process Management) with Rules Engines. You had to define the workflow procedure step by step and you also had to define the conditional logic with a set of hard-coded “business rules.”

As an LLM is a “digital brain,” it can serve as a new kind of Business Rules Engine that makes the decision which path to take in an intelligent business process. If the LLM is the “digital brain,” then the orchestration platform is the “digital arms and legs” that execute the decisions of the “digital brain” connecting with the business applications. Making the process work shifts from procedurally defining it (explicit instructions) to now managing it declaratively, which means that the LLM dynamically decides which next process step to take based on implicit instructions. As an analogy, think about shifting from playing soccer yourself to now becoming a coach who teaches the rules but then watches from the sideline. 

To bring these 3 fundamental AI Integrated Enterprise patterns to life, I would therefore like to share a few of our own concrete projects with incredible business outcomes. 

150% growth in reactivated sales opportunities

Example 1 of the first Pattern of “Operationalizing Unstructured Data” is a GenAI powered Reactivation Go-To-Market campaign. The idea is to re-engage with opportunities that were previously “closed lost,” often due to deferring the project. Gen AI crafted the outreach email and results exceeded all other previous campaigns, improving previous results by 150%.

Cut sales call follow-up time from 11 hours to 2 min

Updating Salesforce and quickly crafting follow-up emails requires a high degree of self-discipline for sales reps. What if you had an assistant that recommends Salesforce entries and follow-up emails based on a sales call transcript? This is a second example of the first Pattern of “Operationalizing Unstructured Data.” At Workato, each sales rep has their own assistant (it’s a “digital brain”). That assistant is wired directly into our Salesforce instance and email system to take actions once the rep confirms. The result is much higher sales productivity and much more consistent information for sales management to act on. Most importantly, this AI Integrated Automation shortened the time to follow up from 11 hours to 2 min, which truly makes a difference to customer experience and efficient operations.

94% accuracy identifying solutions to complex technical issues

For our support organization, we created a Slack interface to semantically query our vast support content and prior customer interactions via an LLM. This is an example for Pattern 2: “AI-Powered Knowledge Management.”

The solution includes the LLM, a vector database and also a mechanism to cleanse content from customer confidential details. The solution is delivering 94% accuracy in resolving the highly complex technical questions asked by customers. This is making our own support staff, especially new team members, much smarter. 

78% more efficient SOAR (Security Orchestration and Remediation) plus AI Power

Our security team automates security responses (both investigations and response actions) with an efficiency improvement of 78%. On top of that, our engineers enhanced the SOAR processes with a novel, declarative process approach. This is an example for Pattern 3: “Smart Business Rules Engine.” Instead of defining the automations step by step as procedures, the OpenAI-powered SOAR solution lets an LLM decide what’s the best next security “task” to perform. The available security tasks (different security incident enrichment and investigation steps as well as security responses, such as blocking a firewall, are defined as low-code automation and the LLM prompt looks almost like a “security analyst job description:”

It’s amazing to see how this declarative approach that creates an intelligent process automation, performs in real life. This will undoubtedly be the most modern, innovative and capable way to automate your security processes. Moreover, this novel approach to implementing intelligent process logic is of course also applicable to many other types of business processes. 

The guide to become an AI Integrated Enterprise

At Workato, we published the “The New Automation Mindset” book outlining the growth, process and scale mindsets that help enterprises transform into AI Integrated companies. 

Growth Mindset

Growth mindset is crucial. You have to be open to new ideas and you have to be ready to bring that same mindset to your business processes. When applying AI and Automation together, by definition, you will not just automate existing tasks but make innovative changes to existing business processes. This is best accomplished by a CEO mandate for continuous process innovation to all operations teams. 

Process Mindset

Many people on the business side don’t naturally think in processes. They also don’t know the possibilities of AI in detail. That’s why the partnership between business and IT is so important. And today, in the “AI Integrated” world, more than ever before, it has to be a partnership instead of the traditional relationship where IT just delivers what the business wants. Gartner was first to spotlight this new partnership across business and IT as a “Fusion Team.” Think of IT as the collaborative technology consultant to the business, the “Business Technologist.” It’s important to link your business technology team tightly with your operations teams so that they can provide active guidance around AI opportunities.

Scale Mindset

Last, but not least, becoming an AI Integrated Enterprise has to be a democratic process. It will only work if as many people as possible within IT and business operations are involved in making the goal happen. Such a pervasive approach across the entire company of course requires smart governance. But an AI Integrated Enterprise effort will be severely bottlenecked if it’s limited to a small centralized team of experts that operates in an ivory tower. A good way to institutionalize this approach is with internal hackathons as well as company-wide process improvement brainstorms. We see lots of process improvement ideas from RevOps teams. The AI Integrated Enterprise needs everyone’s input. 

When an enterprise takes all 3 mindsets - growth, process and scale - equally seriously, this maximizes the success factors to become a lasting AI Integrated Enterprise.

The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.

Other Topics

Technology Innovation