In recent years, machine learning has become prevalent in almost every industry. Machine learning has the ability to learn automatically and bring about changes and improvements from experiences. However, most of the businesses today are facing a lot of hurdles while implementing machine learning projects. For those companies in this space, several organizations are offering support for effective machine learning implementation.
Quickpath is one such company that is proving itself quite realistic by offering AI and machine learning solutions where businesses can take benefits. In an exclusive interview with Analytics Insight, Quickpath CEO Alex Fly outlines how machine learning and AI enable businesses to make automated, intelligent decisions faster.
With what mission and objectives, the company was set up? In short, tell us about your journey since the inception of the company?
I’ve been working in AI for the past 20 years, and previous to starting Quickpath, my co-founder, Trent McDaniel, and I worked in a consulting capacity. We realized that many businesses faced similar challenges when implementing machine learning projects, so we wanted to create software that could facilitate this process without requiring custom solutions each time. Companies are frustrated by the complexity of using machine learning to make business decisions. We wanted to help them to get models out of the lab and into production so that they could finally realize value using the technology. Therefore, we founded Quickpath, which enables businesses to make automated, intelligent decisions using machine learning and artificial intelligence.
Kindly mention some of the major challenges the company has faced till now.
While it’s now easier to build quality machine learning (ML) models with popular open source frameworks, AutoML, and citizen data science tools, enterprises continue to struggle with the difficulty of integrating and managing ML to drive real business value. Industry-wide, only 13% of models built are ever used in production decisions (i.e., analytic shelfware). Production implementations average 6-9 months, cost several $100Ks, and the resulting analytic and technical debt taxes future productivity. Left unchanged, the majority of ML initiatives will fail to deliver value.
The reasons for these challenges from existing approaches span the four pillars of organizational success:
People: Lack of critical skilled resources in key data science and data engineering roles within IT. Lack of understanding of what’s possible and not possible by C-level stakeholders.
Process: Lack of defined, repeatable process for building and deploying AI applications results in heavy analytic and technical debt.
Technology: Rapidly changing landscape of tools and ML frameworks is hard for enterprises to keep up with and to select the right ones.
Information: While there is lots of data in enterprise organizations, it’s still siloed and not well catalogued for repeatable use by a data scientist for model development.
Please brief us about the products/services/solutions you provide to your customers and how do they get value out of it.
Quickpath makes it simple for citizen integrators to build, deploy, and manage production ML enabled apps. Our ever-growing connectors provide the data and ML fabric for companies to tie together first- and third-party data sources, machine learning frameworks, graph databases, and SaaS applications for intelligent decision automation. The platform’s low code design studio and open architecture enable the self-service creation of intelligent decision APIs data, models, business rules, and feedback loops. Our unique approach reduces cost and implementation time up to 90% while ensuring confidence and transparency in ML decision-making with model management and patent-pending drift, anomaly, and bias detection.
In addition to our software platform, we also have a customer success organization that helps our customer to define and build a data science factory within their organization. Our two main customer success offerings are focused on data modernization and data science modernization because successful AI can only be achieved at scale when built upon a solid data foundation.
How is the Big Data/AI/Robotics industry changing? What are some of the key technology transformations in this space?
The pace of innovation and change in the AI space is moving at light speed. There are new scoring libraries and frameworks released daily. The popularity of data processing and scoring frameworks grows and wains almost as quickly, making it challenging for organizations and resources to keep up or even decide upon a set of frameworks that they plan to use for exploratory and production applications. Quickpath’s ever-growing list of connectors and support for scoring frameworks eliminate the risks and challenges by making it easy to support multiple frameworks, pilot new ones using A/B testing, and by abstracting the infrastructure and dependency management of using different data sources, scoring frameworks, and SaaS applications that require integration.
What is the reason that organisations are using IoT/analytics/big data/AI/ML/Big Data Analytics?
Stephen Hawking defined intelligence as “the ability to adapt to change,” which really well sums up ML and AI value to the enterprise when compared to the manual or rule-based processing that it aims to improve. Manual and rule-based approaches to customer interactions and process automation have failed to scale with the ever-growing increase in digital, mobile, and IoT signals and wealth of customer data that companies now possess. Machine learning excels at taking all of these rich signals, detecting patterns within them, and providing algorithms to optimize outcomes to the desired target that would have been impossible using previously available techniques.
Do you also feel that the right kind of talent is a challenge in the industry?
Absolutely. The velocity of innovation and change in the data and analytics space at this time is unlike anything we’ve ever seen before. There are new open-sourced and startup solutions released daily, and the technology stack that a machine learning engineer has to be proficient at is entirely overwhelming. It takes a small army of highly specialized skill sets to take a machine learning model and operationalize it into a production environment: data engineers, data scientists, ML engineers, DevOps engineers, infrastructure engineers, SRE engineers. With Quickpath, we leverage all of the same best of breed technologies behind the scenes, but automate and abstract the complexities of them so that a citizen integrator can do the same work more efficiently and in a much more manageable way than that whole team of the scare, specialized resources.
Could you highlight your company’s recent innovations in the AI/ML/Analytics space?
We’ve led the enablement of a data science factory model at one of our Fortune 500 Financial Services customers where we created a highly repeatable and easily managed path to production for ML enabled applications. We were able to increase their analytic throughput from ideation to production by 16X, resulting in nearly 100 real-time ML enabled decisions a year for a relatively small team.
Another recent example of the productivity gains that our platform offers for data and ML engineering was for a client looking to use a series of Natural Language Processing (NLP) algorithms to better understand and react to conversations occurring within their customer contact center. One of our top data engineers spent 6 weeks working to enable custom-coded data and scoring pipelines to perform speech to text translation, speaker/party identification, tone and sentiment analysis, and topic analysis. Leveraging our platform as a proof-of-concept, a far less experienced resource was able to recreate all of the custom developed functionality in under 30 minutes.