• Written by Muhammed Abdulla NC, CEO

When building a system like AsscherAi, which lets a large organisation query its own data in natural language and get answers in real time, the question of whether to use an agent or a fixed workflow is a critical one. In theory, an agent should be able to handle complex queries and adapt to changing data, while a fixed workflow should provide a straightforward and efficient way to process routine queries. However, in practice, the choice between these two options is not always clear-cut, and the wrong decision can lead to significant problems down the line.

What each option genuinely commits you to

An agent-based system commits you to a significant investment in natural language processing and machine learning, as well as ongoing maintenance and updates to ensure the agent remains effective. This can be a major undertaking, requiring significant resources and expertise. On the other hand, a fixed workflow commits you to a more rigid structure, which can be less flexible and more difficult to change if requirements shift. However, it also provides a clear and predictable process for handling queries, which can be easier to manage and maintain.

For example, a shift supervisor using AsscherAi to check the maintenance log may find that an agent-based system is able to provide more detailed and nuanced answers, but requires more setup and configuration. In contrast, a fixed workflow may provide quicker and more straightforward answers, but may not be able to handle more complex queries.

The case where the other one is right

There are certainly cases where an agent-based system is the better choice. For instance, if the organisation has a highly variable or unpredictable query pattern, an agent may be able to adapt and provide more effective answers. Similarly, if the organisation has a large and complex dataset, an agent may be able to navigate and provide insights that a fixed workflow cannot. However, these cases are often more rare than people think, and the benefits of an agent-based system may not outweigh the added complexity and cost.

In general, people tend to overestimate the benefits of agent-based systems and underestimate the complexity and cost. They are wrong to assume that an agent-based system will always be the better choice, and should carefully consider the specific needs and requirements of their organisation before making a decision.

The cost that only shows up later

One of the biggest costs of an agent-based system is the ongoing maintenance and updates required to keep the agent effective. This can include retraining the agent on new data, updating the natural language processing algorithms, and ensuring the agent remains compatible with changing systems and infrastructure. These costs can be significant, and may not be immediately apparent when first implementing the system. In contrast, a fixed workflow may require less ongoing maintenance, but may require more upfront investment in design and configuration.

For organisations considering implementing AsscherAi, it's essential to factor in these ongoing costs and consider whether they have the resources and expertise to support an agent-based system. If not, a fixed workflow may be a more practical and cost-effective option. You can learn more about the costs and benefits of each option on our website.

A way to decide without a six month evaluation

So how can organisations decide between an agent-based system and a fixed workflow without conducting a lengthy and expensive evaluation? One approach is to start by identifying the most common and critical query patterns, and designing a fixed workflow to handle these. This can provide a quick and effective way to handle routine queries, and can help to identify areas where an agent-based system may be more beneficial. From there, organisations can incrementally add agent-based capabilities to handle more complex or variable queries.

This approach can help to mitigate the risks and costs associated with implementing an agent-based system, and can provide a more gradual and incremental path to implementation. Organisations can also contact us for more information and guidance on how to get started.

What this does not do

It's essential to note that the choice between an agent-based system and a fixed workflow is not a one-size-fits-all solution. There may be cases where neither option is suitable, or where a hybrid approach is required. For instance, organisations with highly sensitive or regulated data may require a more customised and secure solution. In these cases, a more tailored approach may be necessary, and organisations should be cautious of trying to force-fit a solution that does not meet their specific needs.

In general, the key to success is to carefully consider the specific requirements and constraints of the organisation, and to choose a solution that is tailored to these needs. By taking a thoughtful and incremental approach, organisations can implement a system that provides effective and efficient query handling, without breaking the bank or introducing unnecessary complexity.