(US & Canada) | We Help Organizations Rapidly Co-innovate AI Solutions With End-users — Tredence Co-Founder and Chief Revenue Officer

Shashank Dubey, Co-Founder and Chief Revenue Officer at Tredence, speaks with Robert Lutton, VP at Sandhill Consultants, and Editorial Board Vice Chair at CDO Magazine, in a video interview about the last mile problem in AI, and how Atom.ai enables organizations to innovate rapidly.

Tredence is a data science company that provides data analytics and AI services to hyperscale enterprises.

While many organizations may be rich in data and generate insights from it, they may not be able to turn insights into action, says Dubey. This inability to translate insight into action is defined as the last-mile problem.

Delving deeper, Dubey states that the problem occurs for two main reasons. First, the people who actually act on the insights are different from the ones sponsoring the AI solution.

For instance, it is the C-Suite that sponsors the AI solutions; however, the solutions will be used by people running the warehouse and their likes. In this case, the actual users are not involved in the conversation, which creates issues regarding what is needed on the ground versus what is discussed in boardrooms.

Dubey further states that it is critical to build that connection and understand that people taking sales orders do not need an education in math and AI. Rather, they need AI and math to be effortless to make them look smarter, not dumber.

Therefore, Dubey recommends understanding the business processes apart from understanding the end-user. He notes that many times it is about re-engineering the business process and also human change management to ensure that AI is not replacing humans.

According to Dubey, the last mile problem is not the problem of not having the right math or the right AI; rather, it is about not appreciating the human dimension of change management and the end user.

To address the last-mile problem, he formulated Atom.AI. Explaining how it fits in, Dubey discusses a problem statement and the current state of organizational experimentation. He states that the current state is that proof of concept takes months, and rapid prototyping is a big challenge for AI and ML programs.

Then, a quick start remains a distant dream despite investing millions of dollars in tools and platforms; in addition, there is a lack of usability and complexity in the infrastructure setup. These multiple challenges have slowed down the rapid experimentation and prototyping that organizations need to solve the last-mile problem and create accelerated delivery.

Keeping these in mind, Atom.ai, through its multiple accelerators, has helped to reduce the problem discovery time by about 75%. Further, it can reduce the effort around data architecture and design for ingesting, processing, and storing data by up to 80% through pre-built organizational data models and data pipelines.

As far as ML model development is concerned, the company has seen efficiency gains of up to 40% with Atom.ai. It is capable of end-to-end feature engineering and integrated ML operations. Additionally, Atom.ai has aided in faster infrastructure deployment, delivering efficiencies of up to 30%.

Highlighting the architecture aspect of Atom.ai, Dubey illustrates that it is divided into five parts. The core of this is a single-click automated infrastructure deployment, which may involve Snowflake, AWS, or a combination of both.

It is the core platform to provision workspaces and data sets without innovation teams needing to spend excess time there. Then, there are data works in Atom.ai, comprising hundreds of prebuilt data pipelines, standardized data models, and pre-configured external data sets covering more than a thousand methods.

Next comes AlgoWorks, which has numerous AI/ML notebooks across 30 business use cases, and the last element is the solution work, wherein it has pre-built solutions.

Moving forward, Dubey maintains that Atom.ai enables the organization to give a quick glimpse of a solution to the end user. In the process of decision-making, Atom.ai also involves the end-user to understand their perspective and needs.

Consequently, the end user comes onto the platform along with the design thinking experts and the data science and AI experts, and that facilitates rapid innovation and prototyping.

Atom.ai onboards end-users in the beginning and co-innovates with them, keeping in mind the current workflow and processes and making it smarter using AI and ML, he concludes.

CDO magazine appreciates Shashank Dubey for sharing his insights with our global community.