As business users across almost all departments become more digital themselves – as we have all been encouraging them to do for a decade or more now – they demand more. More connections to make their day to day tasks easier; more integrations to free data from one system, liberating it into multiple others where it can add value. The key goal of most organizations today is to find ways to get more value from their IT systems and their data. Integration is a key way in which they can achieve this.
The demand for IT and developer skills has been a matter of concern for a long time. And as the technology demands of companies everywhere get more complex and sophisticated, that demand is unlikely to ease any time soon. As everyone, everywhere is looking at AI and what value it can bring to an organization, there’s huge potential benefit in this area of developing integrations.
The nature of application and data integrations, however, can be a complex one. From defining tasks, to writing glue code and then to testing and maintenance, there are layers of development needed in order to get to the point of rolling out an integration.
But what if individuals within an organization could orchestrate their own integrations? What if Jane in Finance could single-handedly create an integration between an order processing input and a payments platform? What if Bill in HR could create an integration between email from recruiters and Workday? These one-time integrations represent tremendous value…they are also too potentially voluminous for IT teams to handle. There simply aren’t enough skilled developers to manage that level of demand.
What if we could move past the existing technology of a basic integration co-pilot and establish a full-blown autonomous AI Integration Specialist? Accessible to anyone through a natural language chat interface? That seems like a fantastic idea, but tough reality. Well, there is a pathway to get there, and something we’ve been working on with webMethods AI to make it a reality.
A multi-AI approach
While it’s true that no single AI has the ability to do all of the layers of development needed to code a new integration from scratch, it’s also true that this is not necessary to achieve the end goal. Taking inspiration from tools like Auto GPT, bringing together the right ‘team’ of AIs and uniting them under a simple chat interface can get the task done with great efficiency.
We see the future of AI not as a monolithic landscape, with one superior tool ruling them all. Instead, we see a number of specialist AIs, designed to execute very specific tasks, but with great efficiency and accuracy. This is how businesses are going to realize value from the very broad notion of artificial intelligence. As much as multi-cloud support is a key capability of today’s integrations stacks, multi-AI will become as important to leverage the strengths from various offerings and reduce dependencies.
One of the questions, then, becomes how this varied mix is both created and then used. The answer is through using one AI integration layer to manage the others.
As we’ve found in our experience of using webMethods AI to build integrations, that means having an AI abstraction service to manage a series of AIs. One AI to create a task list, another to go out and find the existing API code, another to write the new code, then another to test and manage. This team of AI ‘engineers’ goes away and does all of these tasks, with a generative AI service such as ChatGPT acting as both the interface for the user and the manager of the project. The AI can be trained to ask key project management questions and interrogate the answers, for instance. All the end user needs to do is be the recipient of the outcome, ask any questions they need to and request edits or updates where needed.
Freedom to integrate
The revolution that generative AI in iPaaS enables is significant. There are two main areas where it will be seen the most. The first is in freedom to operate and integrate for the business. Those employees who see a need to integrate information will be able to action it immediately. The speed with which these AIs can deliver a solution is so quick that you’ll barely have time to grab a coffee while it builds it. The increase in output and productivity from individual employees will be huge – as will the reduction in their stress levels.
The second key benefit is to the productivity of the scarce developer resources in your business. Not only will some work be removed from their to-do lists altogether because it’s done by line of business folks, they can also focus their time more effectively. Some more complex integrations will be significantly more involved and need that expert eye to work with the AI platform to ensure success. But the ‘simple’ integrations can be done in no time. By ‘simple’, we mean those where the time needed is more often in seeking out the APIs from platform creators and getting the briefs from the end users, but the coding itself has a low level of complexity. Of course, more time will be needed to maintain the AI itself – no-code platforms in fact require more sophisticated ‘minders’, despite their surface-level absence of code. Developers can also be freed from a number of essential, but mundane, tasks, such as clearing certain types of technical debt.
The questions about ‘if’ AI will be a tool for business value have certainly turned more now to ‘how’ and ‘when’. The key is in harnessing everything that’s available, not trying to fit a square peg into a round hole with an AI not cut out for the task you need; and not trying to build everything yourself from scratch. AI specialists will be the future of business – making time more productive, employees happier and data more valuable.
