Legito PowerUp 2024 AI and Rule-Based Automation of Documents
Ondra needs no introduction to Legito users, but it’s worth remembering that he founded Legito and his enthusiasm and plans for the product have continued unabated. If you have observed the pace of new features added to Legito, you have seen Ondra’s influence.
Let’s start this report from Ondra’s presentation with his succinct definitions to distinguish AI from rules-based automation:
- Rules-based automation: the software isn’t self-learning, but it always returns correct and consistent results
- AI automation: you provide the data, and you let AI do the work
Ondra says you need to be able to answer ‘yes’ to all three of the following questions to identify a use case suitable for AI:
- Can I accept that AI returns different results in identical situations?
- Can I accept that nobody can tell me why AI returned these particular results?
- Can I accept that, if AI returns the wrong result, there is no quick and easy fix?
- Document automation is about accuracy, accountability and risk-management – AI is impressive but the results are not assured.
- Legito uses AI to create marketing materials because any anomalies from AI create no risk.
- But, for creating contracts or long documents, a few incorrect words could have a serious impact.
- Contracts and legal documents demand precision.
- However, there is a benefit of deploying rules-based automation and AI together.
- Ondra demonstrated three examples of Legito using AI inside a rules-based template to obtain better results.
- The first example was a simple script to extract the details of a party to an agreement and asking ChatGPT to provide background on the counterparty.
- The script is simple: Imagine you are a top procurement specialist and you are asked to summarise information about the following company or individual with name [debtors_name] and address [debtors_address].
- Ondra’s second demonstration showed ChatGPT providing guidance about whether to include an optional clause within a contract.
- The third example showed ChatGPT providing guidance on typical interest rates to be inserted in a loan agreement.
- In all three examples, the human stays in the loop to scrutinise the results.
- If ChatGPT produced unhelpful results, there was no risk of the results undermining the integrity of the agreement which is directed by the rules-based components.
- Users can plug in their own ChatGPT credentials, and example scripts are already available on the Legito website.
Tomas Fiala
Director
Since 2018, we've been working with Legito technology in our Centre of Excellence, which provides document lifecycle management and document automation services across the PwC global network. Establishing a partnership with Legito is proof of the success we made with Legito internally.Explore More Success Stories
Sony Varughese
Chief Technology Officer
Weekly meetings with Legito's Professional Services Team made the implementation and change management process collaborative and efficient - overall, an outstanding vendor partnership!
Tawnya Gilreath
M&A Advisor, Director of Forms
Legito gives us the ability to take away things that create issues. It saves so much time, it is really awesome.Explore More Success Stories
Galya Oster
Emerging Market Trade Finance Specialists
Paul V. Saunders
Chief Innovation Officer and Partner
Nel Pelle
Sales Manager
Using Legito, we reduced our checking tasks from 20-50 minutes down to 5 minutes per single contract.Explore More Success Stories
Jeremy Aber
Software Attorney
Eli Cohen
Chief Operating Officer
Increase Efficiency Now
