How Lytica became a unique analytics company: Part 16
It has been about one year since Lytica began its investment in AI research and development through our newly created Advanced Technology Center. Aside from focusing the company exclusively on supply chain analytics, this was the most significant strategic move that Lytica had made and underpins our R&D agenda for the next 5 years. It also forms the foundation for our new product plans and roll outs.
This was a good decision. Choosing to report on our company’s journey through this transformation was another. Both have moved our company forward in ways that were inconceivable two years ago. Our customer relationships have evolved, resulting in increased engagement around product enhancements and new product ideas. We are being recognized more broadly as an innovative technology company which is great because it aligns with our mission to make good companies more competitive by providing the most trusted e-component market intelligence around pricing, security of supply and compliance. Delivery on the promise of this mission requires innovation and cooperation with all of Lytica’s stakeholders – in particular, our customers. One thing that I know for certain is that we have great customers; the best supply chain people from amongst the leaders in the electronics industry. When you consider the Fortune 500 & Global 500 top 5 players in each market segment using e-components, we have at least 1 (and sometimes 2) customers from each segment of that rarified air. When you take into consideration the workshops that build most of the world’s electronic gear, many of the top EMS companies are using Lytica tools. One metric worth sharing is that our CCE application gets more than 2 million hits per week from users. With all of these positives, I’m still most proud of the fact that nearly 100% of first time customers become repeat clients – a testimony to the large and tangible ROI of our toolset.
Some other signs that we are on the right track include our participation in cutting edge AI research through Federal grant programs, our new university interaction program comprised of partnerships with leading researchers at two universities and seeing the fruits of our internal projects materializing. We leave the first quarter of 2018 with a record month & quarter in sales and begin the second quarter with the exciting announcement that we have carved out Gnowit Inc.’s AI consultancy and service arm to join our company. Gnowit has been our external AI partner and are now officially Lytica’s ATC research team.
With these new researchers joining Lytica in April, I have created a Chief Research Officer position to complement my role as CTO. We will both oversee the activities in development; the CRO will be the prime on ATC research and I will be the prime on commercialization. As our mission and products are about trusted e-component market intelligence, we are focusing our 2018 exploration on machine document reading to capture, decipher and organize information into usable intelligence. There are many kinds of documents pertaining to e-components that supply chain professionals and engineers need to efficiently perform their functions. We made significant strides in 2017 with auto-reading of simpler documents and tables; our 2018 program addresses the more complex challenges – particularly around verification.
While it’s relatively easy to verify a hundred things at a time, it becomes more difficult with a thousand and downright hard with tens of millions. As data is being collected and assessed in enormous qualities, the development of verification techniques to keep up with our AI system output has been given a higher priority. We are using methods such as those from Dr. Edwards Deming wherein we use two different techniques to obtain the information and then compare their results. If each technique has a low error rate, the probability of the same error occurring in each is extremely low. This comparison technique can be used to ensure high quality if the common results are accepted and conflicting ones are investigated. Another advantage of this dual approach is that the non-AI generated data can be used to train the AI system to get better results.
With our 2017 research moving through development to commercialization (our RD&C cycle) we will begin new initiatives involving deep neural networks and natural language processing for next generation products. We will also continue our work on estimation, prediction and automated data error correction. Our 2018 development initiatives include SCM tools to aid the concept and product prototyping function as well as co-developments with customers through our ATC incubator program. 2018 commercialization will bring five new products to market. Next on the list is our BOM Cleanser, which awaits completion of the verification of our MPN & Company Names Library. After all, it takes time to verify millions and millions of data.