Skip to main navigation Skip to content

Solution-Centric AI

Solution-Centric AI - Person writing notes

AI, Machine Learning, Natural Language Processing — these have become some of the biggest buzzwords since blockchain, cloud computing and synergy. Massive investment is being poured into AI initiatives and technologies, yet success stories are rare. Few AI initiatives have left the “experimenting” phase, and just a fraction have made it into production. Most would point to the fact that AI is “cutting edge” technology, despite existing in one form or another for decades.

We don’t buy this explanation. Instead, we believe the factors inhibiting AI adoption and deployment are more human than technological. From our perspective, this is primarily a result of technology being designed and built by computer and data scientists. Alternatively, it emerges from the tools being made ubiquitous by cloud providers for these kinds of users — but cloud providers have one goal, which is to sell more cloud compute and storage resources — not to increase client productivity or deliver unique insight.

This leads to a standard problem in the technology industry — tools and products in search of a problem. Further, it often results in user interface and experience being a superficial part of the project, rather than one foundational to the technology. Finally, it means that even if the technology is capable of providing insight, that insight is rarely accessible to decision-makers and leaders.

Urvin has a fundamentally different approach that we call solution-centric artificial intelligence. Our team has substantial experience in bringing artificial intelligence projects to production, but leverage the technology as a means, not as an end. We have been doing this for decades. Our process is unique in the space, in that we can leverage proprietary intellectual property and deep experience. User interface and experience is foundational to this process and approach, and is incorporated from the beginning of any project.

The solution-centric approach is centered around what we call the “AI Lifecycle.” This process ensures a deep understanding of the challenges that a business is facing, and the resources they have available to address those. It means an integrated client relationship that is not project-based.

AI Lifecycle Data to Insights to Prediction

AI projects are not standalone — they do not exist in a vacuum. We start with a robust data analysis and engineering phase that helps clients understand exactly what data they have, what state it is in, and how to ask the right questions. We partner with our clients to understand the problems that they’re facing, and what kind of solution and experience would help them leverage the value in the data that they have.

We categorize our ongoing efforts into three types of solutions:

  • Discovery: finding insight deep within oceans of data, uncover subtle and latent relationships, and design innovative visualizations for exploring and extracting insight.
  • Detection: combining digital forensics and pattern recognition to create solutions for everything from quality control and fraud detection to image analysis and customer classification.
  • Prediction: understanding how systems are evolving and forecasting where they might be headed, turning unstructured data into structured data to use as an input, and leveraging complex systems analysis to know whether the system is predictable.

Once we start down one of these paths, we can deploy different technology to meet our client’s business objectives — we have powerful proprietary technology and are able to integrate best-of-breed open source technology. Each of the top-level solution types comprises a set of AI architectures which can ultimately be automatically selected and configured according to specific business requirements and available data. Under the hood, a number of Machine Learning algorithms are packaged together to provide an end-to-end analysis for each solution type.

AI Paths - Discovery, Detection, Prediction

It is difficult to pursue this approach when your technology is overly complex. Compound that with the need to assure the application of machine learning techniques is correct, something that many struggle to get right. We put the priority on addressing exactly those things — our proprietary technology is elegantly simple, borne from a deep understanding of the math behind artificial intelligence, and our approach is sound, coming from our team members each having decades of experience deploying solutions to production. That’s how a full-stack solution comes together, from front-end design to scalable “big data” research platforms and client-facing applications, all integrated together cohesively to provide specific solutions of discovery, detection and prediction. If you’re evaluating AI partners, that’s what you need to look for and it’s what we focus on providing.

AI Thoughts + Insights

Thought
When Artificial Intelligence Fails, the Perils of Accuracy
Despite its utility, “system accuracy” is often the wrong metric to focus on and can result in very problematic behaviors and results.
When Artificial Intelligence Fails, the Perils of Accuracy
Case Study
Legal Document Management and Robotic Process Automation
Urvin was approached by a Fortune 500 company whose in-house legal team was struggling with the volume of legal contracts that they had to review.
Legal Document Management and Robotic Process Automation
Case Study
Data Linking for a Noisy, Huge Dataset
Urvin was approached by a firm who needed to process petabytes of noisy data with a reasonable hardware footprint and processing time.
Data Linking for a Noisy, Huge Dataset