A publicly-traded software company with solutions in a regulated space had challenges keeping up with the high volume of legislative changes / judicial rulings occurring across several regulatory jurisdictions.
They had implemented sophisticated content change-detection and capture mechanisms that allowed them to retrieve large volumes of content from websites publishing regulation-related information.
However, they did not have an effective way of identifying which of the content they had captured was useful for their purposes. There was no way to tag / classify the content from a relevance perspective in order to route it or prioritize it for downstream analysis.
As a result, large teams of experts had to open and review every piece of captured content. This meant a lot of time was wasted opening files and reviewing content that was ultimately not relevant to their business, products, or services.
Previous attempts with keywords and Boolean search - and even some AI / NLP - to filter out relevant content were unsuccessful. The client had engaged with multiple global AI technology brands who also failed at detecting the relevant content "signal" from the torrential flow of content "noise".
WeR.ai's patent-pending ImplicitSearch™ machine learning technology is instantaneously trained by reference content to find and assign a score to content "like it".
Using our AI Bloodhound analogy: A new AI Bloodhound puppy is created at the search session-level and is trained on-the-fly by the "scent" of the content provided by the user. The AI Bloodhound is instantaneously trained to detect content having a "scent" that is similar to what it was just trained on.
The client's regulatory experts provided examples of the type of content they needed isolated from all of the information being brought in by their capture mechanisms. WeR.ai was able to show the experts that ImplicitSearch™ could be instantaneously and reliably trained by the reference content so that it could be used to detect what the experts were looking for.
Our solution will involve placing trained ImplicitSearch™ engines in-line with the content capture mechanisms. Engines trained on a particular type of regulation content will assign a high score to implicitly similar content as it "passes by".
But we're not talking about only one engine and only one "scent" of content that can be automatically classified per its relevance.
The client can deploy essentially a limitless number of these ImplicitSearch™ & Classification "filters" on the the torrents of captured content. This is because ImplicitSearch™ engines are freely instantiated at the search session level and we don't throttle the number of ImplicitSearch™ sessions performed.
If users are interested in regulation XYZ content, then they simply start a new search session, provide some examples of what XYZ content looks like, and then deploy that engine within the content capture and storage mechanisms. XYZ content will be given a high AI score by that specific engine, and can be routed or presented using RPA or other basic workflow logic so that the expert has what they were looking for - right at their fingertips.
This AI technology does not have the "traditional AI bias" of other ML approaches. A pristine engine is "biased" only by the user's action when they apply the "scent" of content they want to find more of.
The client is deepening its relationship with WeR.ai for reasons that include the following:
A startup company offers a platform & community that allows members to control and monetize their personal information.
The company wanted to provide an intelligent staffing service to its members. The service would offer a powerful way for members to find jobs for which they were best suited, based on AI scoring of their resumes.
And because their platform is all about personal data monetization - not only is the staffing service going to help them find jobs for which they appear to be a great fit - community members are also going to make money in the job-matching process.
ImplicitSearch™ allows community members to upload their resumes and be told which open positions are a good fit for them. It also helps recruiters find higher-quality candidates in a streamlined, more consistent way.
Essentially, ImplicitSearch™ intelligently matches supply with demand.
An overview of the user experience (on both sides) explains how this works:
This startup wanted to focus on its core competencies and go-to-market strategy. They did not want to hire a massive (and expensive) product science capability with data scientists and engineers and then buy a bunch of AI tools (what we refer to as the AI Status Quo model).
Our client instead wanted to have the AI architecture and engines provided as a subscription service. Which, by the way, is being provided at a price point that's only about 1/10th that of the brand-name technology and services giants - with arguably better Service Level Agreements.
Plus, as production-grade ML technology for unstructured content search & classification, our patent-pending ImplicitSearch™ is unique in the market in its ability to be instantaneously trained to match human resources supply and demand at the resume level.
A Fortune 500 medical devices company faced challenges related to product approvals and recertifications. The issues were rooted in regulations requiring systematic medical literature review and analysis to find clinical evidence regarding the use of their devices in appropriate clinical scenarios. Such clinical evidence reporting is needed for not only initial product approval, but also for ongoing product recertification (potentially requiring that multiple reports be submitted per year, per device). The company employed large teams of Medical Writers who spent countless hours performing keyword searches for relevant medical publications, then filtering and assessing within search results the appropriateness of found clinical evidence for Health Authority reporting.
The process for searching publications – and then annotating relevant results for subsequent use for detailed analysis for evidence usage – was analyzed with clinical leaders and Medical Writers. The main request centered around the notion that “Once I have finally found the type of publication I was looking for, I want to find other publications ‘like it’”. Keyword search is unable to achieve this, and currently available AI search techniques also have limitations (including AI model bias) in their ability to find “articles like this one”. So with this usage paradigm in mind, the data science team endeavored to innovate a new ML search method.
A new technique for ML search was invented that allows users to take one or more exemplary publications and instantly train a domain-agnostic, search-session-specific ML search engine to go find other publications like the exemplar(s). Keywords were used to define what we call the "Search Universe". The Human-in-the-loop feedback (implicit + explicit), within the keyword and Boolean-defined "Search Universe", enabled this breakthrough AI-on-the-fly content discovery technique.
Search result annotators helped organize results for clinical reporting purposes.
The ability to search for content using not just keywords – but using an exemplary piece of content – could enable hard ROI while also making possible other unexpected benefits:
Cost-Savings: By allowing investigators to more quickly find what they are looking for, the medical devices company estimated it could save over $10M USD per year in Medical Writer consultant fees.
Performance & Consistency: Because keywords and complex Boolean strings are not required for “pointing the ImplicitSearch™ engines in the right direction”, inexperienced investigators can be more effective, with search efforts across teams being more consistent.
Expert-Initiated Investigations: Experts can initiate investigative sessions by training an ImplicitSearch™ engine on an exemplar, then handing off that search session to others who are less experienced in the domain. This “flips the script” on this type of work (it’s commonplace for staff or clerks to start an investigation then send their findings to a senior-level expert who determines relevancy to the task).
Discovery Utility: The solution could serve as a literature “Discovery” tool because the content used to train the ImplicitSearch™ engines needn’t be existing publications; investigators could instead use elaborately crafted hypotheses to train the engines – the results from which being seen as “evidence of those hypotheses”.
This success story is yet another example where a client had previously failed when engaging with a global brand-name tech giant (in this case, over a 9-month period and $1M+ in spend ... ouch).
WeR.ai was able to apply its expertise in personalized machine-learning search and accomplish - in only 45 days - what the tech giant and its team of data scientists were unable to do at all over 9 months.
And we did it for only about 5% of the cost of the tech giant.