Mass-adoptable AI for Predictive Operations
Gandhi on Salt March (1930)

Mass-adoptable AI for Predictive Operations

Digitization of manufacturing sector has the potential to boost heavy-industry profit margins by three to five points provided people can make the new technologies work at scale. Unfortunately for industrial businesses, scaling AI initiatives has been challenging. Several experts have talked about AI transformation at large enterprises and the focus has been on vision, strategy, people and culture. Those are important but don’t provide a comprehensive approach and overlook multidimensional factors that are necessary to succeed at digital transformation.  

Informed from creating a thriving startup focused on accelerating AI transformation, working with some of the biggest companies in Oil &Gas, Metal & Mining and Semiconductor sector, we have learnt that transformations don't happen from the inside out - they grow on you from the outside in

Based on 5 years of commercialization work at Falkonry, here are four simple rules to guide transformations - whether you are a large company trying to transform, a small startup wanting to be an agent of change, or a services firm hoping to bridge the gap between these two.

Five rules of transformation

Rule 1: Transformations are catalyzed by mass-adoptable technology.

If it were not for mass-adoptable technology, there would be no transformation. There is no dearth of reference points - language, wheel, printing press, democracy, steam engines, telephone, electricity, peaceful non-cooperation, transistor, calculators, Internet, smart phone are how we think of human development. In most #industrial companies, they already have executives who are very good at execution and have strong technical understanding of their problems. Their executives are quite prescient about this but those who don't have operating experience could easily make mistakes about which technologies are good. Mass-adoptable AI technology doesn't imply simple core tech, but tech where complexity has been tamed into a simple artifact from a beneficiary end-user perspective. Such technology should have the absolute minimum prerequisites such as no data scientists required, no labels required, and no historical data required.

Rule 2: Transformations require ardent advocates, who'll put their reputation on the line.

Transformations require ardent advocates and they are rarely insiders. This is what (social, economic) entrepreneurs, who will risk everything they have, do. Consider Marc Benioff. In 1999, he founded Salesforce with the dream to convert software into a service. For 20 years, he continued to lead the charge from the front and in the process he busted many myths and established SaaS as a meta-category. All the reasons that are often cited as causes of failures in enterprise transformation are basic challenges in human behavior and transformations that matter, succeed despite them. 

Rule 3: Buy-in should occur spontaneously at the broadest level possible.

Consider that non-violent social transformation happened not just because of Mahatma #Gandhi - we are celebrating his 150th anniversary tomorrow - but because of the anonymous marchers who joined him. They had to understand and buy-in because they saw the pain firsthand and understood the methods to address that pain. Not those who newly returned from London with barrister degrees or those who had transplanted themselves in the hopes of applying their culture to an "uncivilized" mass. The mass-adoptable tool of non-violent civil disobedience was a competency growth at the broadest level possible - something like 300 million Indians learned it over 25 years - 1920 to 1945.

Spontaneous buy-in in enterprises arises when every stakeholder sees the transformation as advancing their own professional interests, when they are the new owners of the transformation agents, and when the results are genuinely insightful and valuable.

Rule 4: Actors must produce outcomes at marginal costs.

Accountability cannot be spared in organizations whose entire reason for existence is their discipline. Now, transformations can only happen when there is enough inefficiency that there is an incentive for bold action. Good technologies are able to scale down investment needed and create zero marginal cost of value creation. Being brutally honest about value creation is essential for ending up on the right side of transformation - it is just like developing good habits at an early age. Many large companies are trying to do this and it makes the #AI or #digitalization teams uncomfortable. It is important to demand the most return upfront, even within the first 90 days.

Commit to two years for targeted results with one or more opt out events.

AI grows exponentially more valuable over time but that time doesn't have to be forever. Two years suffice. Now, all sides suffer if strategic planning cycles are shorter than that. It is necessary that both sides are held accountable and that requires exit options at meaningful intervals. I posit that the right exit moments are 90 days and 1 year from start.

In the end, this transformation is so hard that both sides of the story have to hold the other sides brutally honest. We are doing that with our customers and have become the "little engine that could" in the process.

P. S. Most conventional approaches for adopting #AI in #industrial companies sound very much like Viant, Scient, and Accenture from the early days of WWW whereas what really won in the end was a mass-adoptable Website design tool such as MacroMedia Dreamweaver. It didn't do everything that the client wanted but what it did made their project costs far more reasonable and they didn't require hiring a new crew who knew HTML. Everyone and their son started doing this until we reached a point that grandmother too needed a site and, then, new tools like Squarespace and Wix were born. We need a DreamWeaver for #predictiveoperations and I think Falkonry LRS is the answer.

#iiot #predictivemaintenance #cbm #phm #pdm #artificialintelligence #manufacturing #digitalization #predictiveoperations

Very well put, Nikunj. Couldn't agree more on the fact that transformative change requires ardent advocates, both external and from within the organization. Another aspect that I think is an important factor for the adoption of AI is the identification of the correct use case. The use of ML yields maximum benefits when the underlying problem is non-deterministic, but there is a significant amount of diverse, tractable data to analyze. For deterministic problems, more traditional approaches (e.g., rule-based inference engine) can be used. Product managers, however tech-savvy, often falter in selecting the selecting the appropriate problem to be solved. This is where external experts like Falkonry can help make the right decision.

回复

Brilliant piece Nikunj. I can attest to it myself given my experience at Falkonry and how that helped me guide my vision at CuriousAI. Based on our experiences at Falkonry, success has easily lent itself to inclusive and ardent advocates - Period! It does take a Marc Benioff and an Elon Musk to lead a transformation that takes a dream to a mobilized effort fueled by self-belief and conviction of all stakeholders. Loved it when you mentioned “Spontaneous buy-in in enterprises arises when every stakeholder sees the transformation as advancing their own professional interests”. A strong advocate who can articulate value creation at multiple levels, be it senior leadership that can envision operational efficiencies at scale to SMEs and operational staff that value workflow optimizations so that they may be empowered to make previously unimaginable contributions on the floor, can make the difference between a meaningful transformation and a wasted opportunity. Value creation is and should be expected from any large scale transformation but zero marginal cost only manifests itself in the presence of religious and fastidious advocates that can help catalyze the spontaneous buy-in from all stakeholders before value creation bleeds into the cost equation with no relevant upside. A transformation is not a “project” - it is a belief. And unless you are ready to put your “reputation at stake” and be a “Mahatma" to evangelize the movement, genuine and meaningful transformations will remain myths. I have personally had the opportunity to work with some amazing advocates within Falkonry's customer teams and can attest to how critical they are to successful transformations. If they are reading this article they recognize that this article is dedicated to them and are probably nodding their heads in agreementl. But thanks for initiating this thought Nikunj - it has inspired me to follow it up with another article that speaks to making AI mass-adoptable for Predictive Operations.

Jeremy Simon

A proven resource helping industrials reach their full growth potential

5 年

Nikunj, many thanks for your article from insights at Falkonry. As a co-leader of a project within the industrial tech space that incorporated Machine learning as a “tool” to achieve our vision, I can tell you of the challenges faced when running a project like this within a company as it relates to alignment.....meaning both company leadership and stakeholders in the business get behind the initiative and are willing to commit financial and human resources to at least reach first learnings before deciding “stop or continue to invest”. I can fully appreciate how a standard AI tool box can help with transformation process allowing clarity on how this process really works and making for a much simpler implementation . Note: I understand there’s also gray that exists as not all projects can necessarily be solved by “standard” configuration but I would bet many important accomplishments can be made in this way not least of which it allows a company to make a first important step....your company and its clients are living proof!

Naresh Bansal

Chief Financial Officer at Menlo Security - Enterprise SaaS, AI and Security

5 年

Great points Nikunj, advocates are the true heroes who have the courage and vision to take risks.

回复

要查看或添加评论,请登录

Nikunj Mehta的更多文章

  • Falkonry patents Time series GPT

    Falkonry patents Time series GPT

    Falkonry has won allowance of its time series deep learning patent application focusing on the use of convolutional…

    17 条评论
  • Unified Smart Manufacturing Architecture

    Unified Smart Manufacturing Architecture

    Industrial organizations are seeking to exploit insights from all of their different operational data sources such as…

    8 条评论
  • 4 Myths of Maintenance Data

    4 Myths of Maintenance Data

    Many in manufacturing assume that there is a relationship between their maintenance data and their machine data…

    3 条评论
  • Time Series Data + AI + Cloud

    Time Series Data + AI + Cloud

    Water, water everywhere, not a drop to drink. We are really good at producing real-time data but we are really poor at…

    5 条评论
  • From the Valley of Heart's Delight to Industrial AI customers

    From the Valley of Heart's Delight to Industrial AI customers

    In preparation for a Falkonry customer meeting, these interesting notes to self came up. I thought they make for…

    1 条评论
  • Don't be all things to everyone - A lesson learned from GE in Industrial IoT

    Don't be all things to everyone - A lesson learned from GE in Industrial IoT

    (edit: Updated the list of layers to include application enablement) Stacey Higginbotham has covered the Industrial IoT…

    14 条评论
  • Software-Defined Sensors

    Software-Defined Sensors

    Whenever we have tired of owning boxes, we have gone out and bought even more. Only, this time, the new boxes are…

    3 条评论
  • Big Data Can't Daunt Industrialized Society

    Big Data Can't Daunt Industrialized Society

    Since the dawn of the industrial revolution, there has been a relentless push for automation. Each successive wave has…

    7 条评论

社区洞察

其他会员也浏览了