So, AI is all over the media, with some saying AI can do everything.
Is that really completely true?
How will AI impact existing Automations?
Is AI just one thing?
Can GenAI create Automation bots?
ChatGPT delivering a Bot
By entering a request into ChatGPT or equivalent, could an Automation Bot be created?
Yes, and No.
Some AI products are good at generating generic code, so they could generate some core tasks that a Bot will need to perform.
For a full Bot to be created it needs to have appropriate logon information, access to environment information, be able to produce the necessary audit and log information, as well as how it will enable a design process for humans to get involved in the event of something unexpected.
Of course, in theory all of the necessary specification could passed to an AI process that would generate the fully functioning Bot deployment.
The current reality is highlighted by the emergence in the market of array of “Cheat sheets”, “20 Best Prompts”, “The ultimate Guide to Prompts”. These make it clear that a lot of input is required to enable any AI process to generate the outcome that is desired.
The amount of input required to generate a Bot will be significant and may require similar effort to create compared to the development of the Bot directly.
Is AI just one thing?
ChatGPT and OpenAI have been grabbing the headlines but there is an explosion of AI products in the market including Google Gemini, Claude, etc.
There is some merging of functionality, so that products that started in the market as discrete areas, e.g. Language Chatbots, Image chatbots, etc. are staring to offer a broad range of capabilities.
Existing software packages are being enhanced with AI capabilities which of course delivers the AI functionality in the existing context of the software and reduces the need for generic AI solutions.
Replace Automation
An Automation that is deployed and working will be delivering benefits to the business. What could AI offer in such a scenario?
Automation technology can integrate with AI tools, including GenAI products, hence any capability that AI tools could offer could be built into an Automation.
Where there is a new requirement that could be met by an AI tool? Would such an adoption lead to a pressing need to replace existing Automations?
Such a replacement would seem unlikely although technology overlaps are not desirable due to licences and breath of skills that required for support.
With the rapid emergence of AI vendors and products into the market, there is likely to be some consolidation of vendors in the near future as the venture money which has funded the AI explosion looks for results rather than market share. With that outlook, it seems that Automation vendors who have incorporated AI features are more likely to be long term solutions.
Is AI constrained if used in Automations?
Many AI products are being marketed as complete solutions so would they be constrained by the use of Automation?
Automation has been designed to work with any IT application which means that the technology could be used with any AI software which can be used either through a GUI or an API.
There is a business question about whether the usage of AI follows a repetitive set of actions to make its use within an Automation effective?
Perhaps the broader question is what is the business processes that varies so much that it could benefit from using AI with human interaction?
AI enhancing existing software
A lot of AI functionality is available as Open Source software and publicly available Models. Therefore, it is not surprising that AI functionality is being added into existing Automation software products. This enhancement is required for the vendors to stay competitive.
Automation decisions for Business
With the rapidly changing capability and options available for AI, what decisions can a business make?
There are many questions, some of which have been highlighted in this newsletter. Three of the options are:
1. Stay purely with traditional Automation and “Hope” – It is a high risk approach!
2. Choose positive action, accept AI features are the future. Embrace AI technology now and find ways to make it work successfully alongside with existing Automation technology.
3. Adopt a cautious path of research, pilot, testing, learning and verifying the benefits of using the AI features that made available by an existing Automation vendor.
We are pleased to see the rate of enhancement of UiPath to adopt and exploit AI functionality. It remains the technology that we will use to implement Automation for clients.
Article Author
David Martin
Managing Director, Ether Solutions
https://www.ether-solutions.co.uk/
