Despite fears of AI entering our world as a disruptive and sudden force, we are already surrounded by intelligent, autonomous beings. Ubiquitous household products such as Siri and Alexa are examples of Intelligent Agents already embraced by consumers, as well as robot vacuums that can move around rooms by their own accord.
But the consumer world is just the first step. Taking Intelligent Agents to the business level promises huge benefits, and we have leveraged this technology in some of our recent engagements. In this blog, I discuss the growing phenomenon of Intelligent Agents, and highlight use cases for their capabilities in the business world to create unprecedented benefits both for customers and internal business operations.
What is an Intelligent Agent?
Though we may be familiar with the products above, the actual definition of an Intelligent Agent is broader. An Intelligent Agent is a program that can make decisions or perform a service based on its environment, user input and experiences. These agents make use of cutting edge Artificial Intelligence (AI), as well as other technologies such as Machine Learning (ML) and rules engines.
Intelligent Agents observe their environment through sensors. This gives them an idea about what the world currently ‘looks’ like, so when a change occurs they can observe it, make a decision on how to act, take the action, and then observe and learn from the outcome. For example, Siri or Alexa will observe sensory input in the form of voice commands (using sensors), search the internet for the required results, decide which results to choose, and present the information back to the user through a speaker (the ‘effector’). The user can also make commands such as ‘play music’ or ‘book travel’, which sets off a similar process. Siri and Alexa are then designed to learn and adapt as they encounter new information and get to know their users’ preferences.
This same cycle occurs for Intelligent Agents used in a business context. For example, there may be an agent that observes a budget change on a project, via stimuli such as cloud-based events. The agent is then able to make an appropriate decision on how to act based on pre-set rules, as well as reinforced models from previous decisions. In this case, the agent’s rules engine or previous experiences may specify that a budget change requires further approval, prompting the agent to send an approval email to the relevant person. Over time, the agent may learn through feedback that for budget changes of less than a certain amount, further approval is not necessary, and so will stop sending emails for cases that fit this criteria.
A different example might occur in a more physical context – a customer arrives at a physical location such as an airport or shop. The Intelligent Agent may observe their arrival via device changes (eg. mobile phone proximity) and decide to notify the customer of an offer via a mobile notification or text.
By taking action and introducing changes to the external environment, the agent creates new stimuli that can be picked up by other agents who may then make changes and so on. The original agent too can observe the results and reinforce the decision in the same way (some) humans learn from their mistakes or past successes. This ability to generate new stimuli demonstrates the vast potential for Intelligent Agents across an entire business or product lifecycle.
Four Ways that Intelligent Agents Drive Business Benefits
Intelligent Agents can be applied in many areas of a business. Below is an overview of some of the broader benefits and real world examples of where we’ve applied this technology.
Personalised Customer Experience
Businesses have more data on their customers than ever before but that doesn’t mean creating personalised campaigns is an easy process. Intelligent Agents are a great way to present contextual offers to customers without a lot of complicated data science modelling. It could be as simple as pushing a notification offering a passenger a free VIP lounge pass as they arrive at the airport. These types of activities can make a customer’s physical interaction with a brand more enjoyable and personalised, and will ultimately help to increase loyalty and repeat business.
Streamlined, Dynamic Governance
Intelligent Agents can remove the need for lengthy governance documents by enabling digitised, dynamic governance with checkpoints. At a construction company we’re working with, Intelligent Agents are being used to increase or decrease governance based on certain parameters. For example, there may be a bid proposing a 40 storey building be constructed in Location A within 6 months, using two cranes. The agent will be able to observe this and create a checkpoint for someone else to determine if the bid needs to include three cranes. Or, if the risk position of the project were to change, the solution would use predefined rules (and any past experience) to increase the governance and route approval requests to the right people. In this case, C-Level or higher managers are only notified when necessary, keeping governance from becoming too burdensome but also ensuring the correct approval channels are followed. The end result is lower cost governance with better risk resilience.
Intelligent Agents (like all ML solutions) keep track of all of the decisions they make, which can be used both as a source of continual learning and as a digital audit trail. Naturally, this makes compliance methodology more objective in comparison to human decision-makers. The information gathered can also be used to check, validate or make changes to business operations, procedures and processes, ensuring the best possible outcomes based on previous data.
Intelligent Agents can also have huge impacts on operational efficiency, imitating human decision makers with greater efficiency and integrating with various other forms of technology. For example, we have applied intelligent agents to optimise the flight loading process for a major freight organisation. The agents are able to observe loading events and demand for various categories of freight and optimise the allocation to different flights. The agent is aware of and can learn how to do this based on business drivers and logistical constraints, for example that animals cannot be flown with dangerous cargo such as dry ice.
We are also implementing Intelligent Agents to help reduce traffic congestion and encourage more streamlined deliveries around a port. In this case, the agents observe the traffic around the port and integrate with a mobile application used by truck drivers. The agent tracks when traffic is highly congested, as well as when trucks enter a particular geofence close to the port. If the traffic is highly congested when a truck enters the area, the Intelligent Agent can encourage the driver via a mobile notification to deviate or take a different route.
Intelligent Agents are an incredibly flexible and accessible technology because they are an architecturally-led rather than vendor-led solution. They can be delivered as standalone solutions or used as a component in a larger set of technologies. For example, in the construction company referenced above, the solution utilises Intelligent Agents deployed on AWS infrastructure, embedded into Microsoft Teams using the Fluent UI Framework. This is a great example of the interoperability of an architecture-led approach to Intelligent Agents.
Intelligent Agents can be considered as a component in Hyperautomation, working with technologies such as RPA and Business Process Automation to enable integration, transformation, and BAU business processes. Leveraging cloud-based serverless technologies is allowing Intelligent Agents to play a larger role in complex use cases where data or events (stimuli) can scale dramatically.
As an architecture-led solution, Intelligent Agents come with a number of foundational building blocks to deliver the required outcomes. Some of these could sit within an organisation’s existing technology ecosystem, and Intelligent Pathways can also bring core accelerator assets such as Policy Engines to streamline logic development. What’s key though is to ensure the underlying set of rules are accessible to the stakeholders responsible for the solution.