Question: Can you really improve your Business Process Orchestration with AI? Has the technology advanced enough to make a meaningful impact?

Answer: In short, yes, and during this post we will elaborate more. All of this is based upon hard won experience DMS has gained doing business process improvement work with our clients.

At DMS the major focus of our work is on systems integration and business process orchestration. We’ve long since used AI to help us with activities like code generation, and now we are actively using AI to supplement our work in the area of Business Process Orchestration.

Readers will likely have heard talk of Agents, “Agentic AI”, the “Agentification” of software services and the “Agentic Enterprise”. It is tempting to dismiss this as industry hype but we can attest that while it maybe nascent this is very real and we among others are using it today. Agents are a great way to start to apply AI improvement to your business processes.

We have observed three types of agents approaches emerging that can help you improve your Business Process Orchestration.

1. Knowledge Agents:

What are they?

These agents have been trained and configured to understand a business process. Users can interact with them to validate steps and explain details. E.g. What is the 123 form and who completes it? Where is it sent to and can you share complete example? How is the calculation done for Box 4a? Etc.

How can they help?

These agents can “learn” complex and involved business processes and then quickly provide detailed information to users who are tasked with completing these processes as part of their job.

Status: Mainstream, available now & within ability of most information workers.

While you still need to invest the effort of documenting your processes, training agents can be fast and their ability to ingest items like flow chart diagrams and then use them to inform querying users is impressive.

2. Action Agents:

What are they?

These agents are evolved knowledge agents that can offer to undertake actions on a user behalf. E.g. Make a booking or system entry, pull and splice system data in a suitable format etc.

How can they help?

These agents aim to save end users time by augmenting and accelerating tasks on the users behalf.

Status: Leading Edge but here. Some scenarios will need input from software developers.

While some information workers will be able to configure a desired action step, the actual undertaking of a task may likely involve the need of some custom code. An example of this is adding skills in Copilot Studio using Microsoft SDK.

3. Autonomous Agents:

What are they?

These agents can pick up unstructured data like email/web form requests, determine the most appropriate business process flow and then undertake execution, potentially escalating to a human user for further approval or guidance as necessary.

How can they help?

This type of agent has exciting potential. Whereas the first two types of agents above tend to ratio at one human to one agent at a time, and focus on improved efficiency; autonomous agents make it possible to scale many agents per single human, driving both efficiency and  scale of the related business process.

Status: Bleeding edge. Doable but requires more investment currently.

At the time of writing, building these is possible but it definitely requires more heavy lifting than most information workers would likely be able to address. Companies like Microsoft are in the process of democratizing further to make this kind of agent more readily available to information workers.

Ok, but how?

If all this interests you and you’d like to target your own business processes, the natural next question is how & which tool(s)? At DMS Group we are predominantly a Microsoft shop, we do a lot of Power Automate and our clients are managing secure access of their data using Entra ID (AAD) access. If that’s you then we’d recommend you consider Microsoft Copilot Studio as a good starting point. Tooling like Azure AI Foundry provides notably more granular control if needed but our experience to date shows that for many business processes this is unnecessary. Furthermore aligning around Copilot Studio for Business Process work keeps more control where you want it: with the information workers driving your business processes.

If you’d like help to accelerate your business process augmentation with AI  tooling reach out to DMS Group – we’re here to help your business achieve more.

Footnote:

This article assumes some rudimentary AI knowledge from the reader. At the time of writing this is still a nascent area of work, others may confer different taxonomies to Agent types both beyond Business Process Orchestration and as this technology space evolves. Copilot Studio, Azure AI Foundry and Power Automate are Microsoft trademarks mentioned with permission.