Avinash Misra is the CEO and co-founder of Skan. Avinash is a lifelong entrepreneur with a proven record of taking ventures from seed to liquidity. He has built successful ventures in the enterprise digital transformation space and his last venture was acquired by Genpact (NYSE : G). Avinash’s insight for Skan took seed in large scale Business Process Transformation projects which he has led over the last decade.
Your previous company Endeavour Software Technologies was eventually acquired by Genpact. What was this company and what were some of the key lessons that you learned?
This company was a front-office digital transformation specialist. That is, it specialized in the build and deployment of specific technologies such as computer vision, chatbots/ natural language processing (NLP), and enterprise mobile apps to improve and transform customer-facing business processes.
We learned two key lessons. First, when technology is applied for its sake only, it creates both technical and process debt. Second, the most value is derived when technology specifically approaches the end user with empathy and a design-think mindset.
Could you share the genesis story behind Skan?
“Automation begins when automation fails.” In one sentence, this was our beginning. When we built RPA bots for complex business processes, we repeatedly noticed that once a bot was deployed it failed quickly because it did not take into account all of the nuances, permutation, and exceptions of that business process. Every time a bot failed, it became one more missing permutation of work. It was an endless cycle of deployment and failures.
So, why don’t we know all the nuances of business processes?
We don’t know all the nuances of business processes because all process discovery is done by human business analysts who ask the process agents to describe work. Humans are spectacularly unreliable in describing things that have a sense of familiarity or habitual and routine. These are often things they can do well, but can never describe with the needed accuracy. Hence, we built Skan to observe real work and understand that work and the processes, rather than interview and document humans.
Skan is partially a process discovery platform. Could you define what process discovery is for our readers?
Process discovery is a broad term that refers to the act of discovering or learning how processes work at an operational or structural level. This is particularly challenging with processes that involve human-system interactions with hundreds or thousands of workers, dozens of software applications, and complex workflows. A great example is the claims management process.
Today, Skan is actually more than a process discovery platform. Skan generates a deep understanding of work (process discovery) and provides advanced analytics to help process owners and transformation leaders measure, analyze, and improve KPIs that drive business outcomes such as the customer experience, revenue, and cost. We call this broader capability: process Intelligence or the systematic collection of data and the end-to-end process and application of that knowledge to control business outcomes or to learn, understand, and make decisions.
According to a study conducted by Ernst & Young, 30% to 50% of automation projects fail. Why do you believe this is so high?
Based on working with our customers, we find that one of the key obstacles to automation success is lack of visibility into current state of KPIs across the lifecycle of automation projects.
For instance, in order to qualify an automation project, we need to baseline the current state KPIs and build a business case. In the experimentation phase, we need to identify technology patterns and define target (to-be) KPIs based on current state KPIs. During the design, develop, test, and operationalization phase, we need to align with the root cause of the problem to solve.
Finally, in the validation phase where we measure investment payback and benefits realization, we need traceability to the to-be KPIs. So, we see that across this entire lifecycle, transparency and traceability to current state KPIs and root causes is required. And, yet, according to Forrester Research (2021), only 16% of organizations say they have complete visibility into how processes work. It’s no wonder automation projects struggle to deliver value.
Can you explain what procedures Skan takes to protect the privacy of people that are being monitored and sensitive business data?
It is important to note that we do not monitor people. We only observe specific elements of work (not the whole screen). These elements are specific work applications that are predefined upfront.
That said, for any applications observed, all sensitive work data is redacted. We also have the ability to anonymize the link between the person who did the job and the process. The names of individuals working in the process can be anonymized, too.
Could you discuss how Skan uses machine learning and specifically deep learning?
Skan incorporates several AI and machine learning algorithms to address various problems such as anonymizing sensitive information (both text and image data), abstracting low-level events to business activities, inferring process graphs, and discovering process variations.
What are some examples of actionable insights that have been gained from this process?
Skan helps process owners and transformation leaders measure, analyze, and improve KPIs that drive business outcomes. Some example insights are:
- Unit cost of production
- Resource (workforce) utilization
- NPS improvement
- Automation discovery
- First pass rate
- Process compliance
- Capacity (workforce) planning
- Reduced process variability
What’s your vision for the future of process intelligence?
Our vision for the future of process intelligence is to transform the way people work so they can improve productivity and reach their full potential.
Today, the global pyramid of work has a broad base of non-value added tasks and a very narrow top of value-adding tasks. Our vision is for process discovery to invert this pyramid.
Thank you for the great interview, readers who wish to learn more should visit Skan.
- The Black Box Problem in LLMs: Challenges and Emerging Solutions
- Alex Ratner, CEO & Co-Founder of Snorkel AI – Interview Series
- Circleboom Review: The Best AI-Powered Social Media Tool?
- Stable Video Diffusion: Latent Video Diffusion Models to Large Datasets
- Donny White, CEO & Co-Founder of Satisfi Labs – Interview Series