Speed and agility trump perfection in today’s hyper-competitive and highly regulated business world
I’ll wait till the next version thanks…
I am constantly struck by how people have a skewed take on accuracy, efficiency and what is “good enough”. In my experience, we all too often reject the offer of a good or service that will provide real benefit now whilst waiting for the nirvana of 100% believing erroneously that anything less than perfect is a risk. In fact, we are losing benefits whilst we wait for the perfect moment. Further, we over-estimate the accuracy and efficiency of existing processes so maintaining the less than effective status quo.
Not perfect, but look at the alternative
There is no such thing as perfection in software – we are constantly looking to improve and add capability. Even devices we rely on like cell phones & SatNav’s are driven by software, which occasionally hangs or lead us to the wrong destination. As consumers, we accept that these anomalies happen and unless they become a regular occurrence, we generally put up with them.
We don’t enjoy 100% accuracy, or you might say ‘perfection’, but we continue to use them because the advantages far outweigh the disadvantages.
Business is different
When looking at the field Seal is in — AI-based Contract Discovery and Extraction, there is still a feeling that until 100% accuracy is guaranteed, there will always be some reluctance from the mid-to-late adopters to wholeheartedly embrace such technological advances. The early adopters know that to achieve 100% accuracy requires a willingness to train the software to become more accurate until that nirvana is reached. That’s the point of AI – train the system to think for itself. The more effort you put it initially, the more dynamic the system can become and the closer you get to that nirvana.
An Example: IFRS 16
Imagine that due to a new regulation, say the upcoming IFRS 16 accounting change, I need to find all my contracts that might have reference to commercial leases, work out which ones I need to review, and from there, move to a program of remediation and review. Remember, the discovery and extraction is not the project – it’s the pre-cursor to the real project which is working to ensure you are compliant with the new regulation.
Exhibit A: We do it this way over here
Historically an organisation would dig out their contracts and engage with a Legal Process Outsourcing (LPO) firm to read them and then manually update a spreadsheet. Leaving aside the time taken and the cost, the assumption made is that because of the human touch this will yield a highly accurate result. And if anyone has ever spent hours, days or weeks reviewing contracts, you will know how the mind starts to wander and quality decreases. Also, humans perceive things differently, and text and concepts subjectively, so errors are part of human reviews.
Exhibit B: The new way
Now, when proposing an alternative solution comprising automated contract discovery and extraction based on several technologies that fall under the “AI” banner, the discussion always turns to “how accurate is it?” Of course, sometimes the results are very, very good, but by the nature of what we do, sometimes the accuracy is less.
But that misses the point.
The question should not be “how accurate is it? Instead, the question should be: “Will using this technology decrease the cost and time taken for my business to get to a fully completed spreadsheet so I can start the remediation work?”
And the answer is a resounding yes. Then you have the capability to refine analysis and train the engine to become ever more accurate.
Recent real world projects have shown that combining automated discovery and extraction with effective process and manual review dramatically reduces the time needed for extraction.
This means you can get started on the real work faster, and at the end of the day, isn’t that the point — speed and agility triumphing over seeking nirvana from the get-go.