Is it Finally Time for Predictive Early Case Assessment (ECA)?

Most people working in the legal industry are aware of the importance of Early Case Assessment (ECA) in litigation preparedness. ECA is broadly defined as the process of estimating the risk (including time and estimated spend) involved in defending or prosecuting a legal case based on the available evidence (data) and other factors. The current ECA process involves the awareness of possible litigation and quickly:

• Determining the basis of the case
• Determining the timeline and potential custodians involved
• Developing a content/keyword list

Collecting/reviewing as much relevant data as possible to help determine potential risk, defense costs, and go-forward strategy, i.e., go to trial or settle

The ECA process varies from organization to organization, but at its most basic level, the ECA process helps corporate legal departments and their law firms determine:

1. If the case defendable/winnable?
2. If they have the relevant data to support the case?
3. If they have/should have anticipated this case and if so, when?
4. Potential keywords and phrases
5. Depending on the chosen go-forward strategy, what could be the possible consequences of going to trial or settling; i.e., corporate reputation, personal reputations, the effect on shareholder equity, and potential release of IP and corporate know-how?
6. Approximately how much it would cost to defend? Is the estimated cost to defend less than or greater than a possible settlement?
7. How much time would be involved?

The ECA process is usually a reactionary process – Company A is suing my company for IP theft - what data do we have that could explain this action and what should our go-forward strategy be, how much will it cost to defend, what could be the downside?

However, ECA should also be an anticipatory exercise as well. For example, while attending a trade conference, an employee overhears talk from a third party that a competitor is thinking of suing because of employee poaching. Would this be considered “reasonable anticipation of future legal action?” What if that story showed up in a blog?

Without proper anticipation or notice of legal action, companies can’t know what to look for to begin an ECA process - until it becomes obvious. However, questions of when a company should have been able to anticipate future litigation have cost many companies large sums of money and in many cases, loss of the case.

But does it need to always be reactionary? Instead, could it be preemptive?

AI and ECA

One of the more difficult aspects of anticipating future litigation is being aware of and actively looking for possible triggers i.e., a business partner files suit against one of your competitors, or a large consumer products company sues a company for unauthorized use of IP– IP your company could be utilizing, or a competitor is quoted in an industry news story saying they are thinking of suing many of their competitors for unauthorized use of their intellectual property.

Wouldn’t it be great if someone (or something) tapped you on the shoulder and said: “you should take a look at this…”

What if?

What if an Artificial Intelligence (AI) continuously monitored business news streams (audio/video/written), as well as other activities within the organization and was able to determine if the companies and content mentioned in the news streams could legally affect your company? What could that AI do to reduce your company’s risk and potentially provide you with a legal advantage? For example, an alert for possible future litigation based on trending industry news, or the automatic placement of an early litigation holds ensuring against spoliation.

Email/communications indications

Following that what-if scenario, once an alert for potential litigation appeared, the AI could immediately scan live email accounts, email archives, and other communications channels to see if employees had been in contact with and were doing business with the subject companies and if so, which employees were in communication with them. Additionally, the AI could scan published articles and map content keywords and even sentiment to determine if further investigation were needed and a content subject heat map could be created to help determine the level of industry focus over time by the press.

Work documents in file shares and cloud storage

If there had been contact between the subject company and yours, the AI would next scan work documents for potentially relevant content in the affected custodian file shares, cloud accounts (OneDrive), SharePoint content, etc. to further determine the level of contact and potential information sharing.

Based on predetermined policies, if monitored subject matter did meet ECA trigger policy guidelines, the legal department would be automatically notified and sent detailed reports which could include:

• The potential matter and legal issues
• A rating of potential liability
• A listing of potential content and overall data stats – the amount of data, file types, age-range – and a current data map
• A listing of possible related custodians and their positions
• A communication relationship map, timelines, heat maps, trends reports, sentiment analysis, key concepts cloud, possible smoking guns, and projected cost to defend
• A listing of documents/email that could be relevant with links and descriptions of where they reside
• The ability to sample review the potentially relevant content
• The ability to automatically copy all relevant content (in a legally defensible manner) to a secure repository for legal hold and review, or apply an in-place legal hold
• Indications if there are any obvious holes in the data, i.e., potentially missing data - responses, deletions, large holes in the communications timeline
• Suggested documents to be used for further AI/ML learning

Naturally, this AI-driven solution would potentially act as the company’s notice for potential litigation so should be taken seriously and reviewed quickly.

Is this blue-sky wishing?

Much of this narrative may seem pretty far-fetched for legal personnel not familiar with AI/Machine Learning or Predictive Coding. The use of AI/ML has been in use in the legal industry for many years now in the form of Predictive Coding, Technology-Assisted Review, Computer Assisted Review, and many more. Back in 2015, Microsoft purchased Equivio, one of the original Predictive Coding eDiscovery vendors, to include in their Office 365 E5 license – advanced eDiscovery.

Predictive Coding has been accepted in the courtroom so it would seem that the next logical step for the legal industry/corporate legal is to harness AI/ML for predictive legal risk monitoring, i.e., predictive early case assessment. But, keep in mind, to effectively utilize predictive ECA, enterprise file consolidation will need to be adopted along with the use of AI/ML to ensure complete data availability.

The AI is greater than the sum of its parts

Back when I worked for Recommind, one of the creators of predictive Coding, we successfully made the argument that training a computer to find/choose potentially responsive documents out of data sets measured in the millions was far more accurate than relying on human reviewers. In fact we found, and convinced courts, that groups of legally trained employees could/did have wildly different interpretations of the same document content depending on numerous variables including where they went to school, what country they were residing in (contract reviewers), if they were married, what they had for dinner the night before, whether they had young kids, etc. On the other hand, algorithm-based computer programs, relying on document test sets of both responsive and unresponsive examples, were far more consistent, accurate, and blindingly fast.

The question posed by a famous/infamous company at the time was; could an AI tell the difference in the meaning of the word “Penguin”, an inside joke for many of us, by understanding the content around the word; for example the difference between the black and white flightless bird, the publishing house, or the comic book villain? It turns out that by using machine learning techniques such as natural language processing (NLP) and others, computers could be trained to understand the meaning of content very accurately.

The next obvious step many are working on is to include AI/ML in the auto-categorization of the millions of documents your company sees daily. What if the employee could be completely removed from the information management equation? How much more valuable would your data be if they were effectively managed?

As I mentioned a couple of paragraphs back, Microsoft purchased this capability back in 2015 and has been expanding it ever since to include into both their Office 365 and Azure Cloud platforms. For those of you that already have Office 365 and Azure Cloud accounts, speaking to your Microsoft rep to see how these new capabilities can take you to the next level in legal preparedness and information management would be well worthwhile.

I can envision the same capabilities for AI/ML when tied to live news streams and company information to recognize potential future legal liabilities, i.e. Predictive ECA.