I have been a fan of Marc Andreessen since the Netscape days — he has consistently predicted the macro changes in numerous marketscapes before virtually anyone else. Recently, I was watching Marc on Youtube “Why You Should Be Optimistic About the Future” and found his discussion on AI particularly enlightening, and in complete alignment with Augmentir’s journey. The entire video is worth watching, but the discussion on AI runs from between the 7:00 to 9:00 minute mark.
Some of the most insightful (paraphrased) quotes include:
- “There is a more fundamental question — is AI a feature or an architecture?”
- “A16z sees this with most start-up pitches now — ‘here are the 5 things my product does…and oh yeah, AI is always bullet number 6.’ Number 6 because it was the bullet they added after they created the deck”
- “If AI is a feature, then this is correct, where every product will have AI sprinkled on it.”
- “We (a16z) believe AI is an Architecture, and if it is, everything above this will need to be rewritten.”
- “Ultimately, the goal of AI is to answer questions, even before the have been posed.”
At Augmentir we had to make a strategic decision at the time of company founding (late 2017), as to whether AI was going to be a feature of our connected worker platform or, whether it was going to be the architecture that our connected worker functionality ran on. We didn’t frame the decision as elegantly as Marc did, but we nevertheless asked, “will AI be a feature of our product or will it be pervasive?”
Even though no one in our space had chosen this path, we decided AI would be pervasive. We postulated that the purpose of a connected worker platform wasn’t to deliver instructions and remote support to a frontline worker, but rather to optimize the performance of the connected worker ecosystem. We knew that AI was uniquely able to address the fundamental macrotrends of growing skills gaps and the loss of tribal knowledge. With an ecosystem of content authors, frontline workers, subject matter experts, operations managers, continuous improvement engineers, and quality specialists, we predicted that there were dozens of opportunities to improve performance.
By building our connected worker platform on an AI architecture, all data is automatically pipelined, labelled, and cleansed, and is immediately available to start generating insights and recommendations. On this journey, the scope of what we can use AI for has even surprised us. Our initial thoughts were on personalizing instructions and content to make each frontline worker perform this current task safely and as quickly as they can, given their current proficiency. This immediately expanded to a generalized True Opportunity™ system that uses AI to stack rank where an organization has the largest capturable opportunities across all stakeholders. The range of this is astounding: which jobs have the largest monthly opportunity, which workers can benefit from targeted training, what is the optimum time to perform any given task, what inline training material can benefit from an update, what content/procedures would benefit the most from an update, etc.
The future looks even more fantastic — AI bots offer a realistic opportunity to capture tribal knowledge and convert it to a scalable corporate asset, and AI Diagnostic bots to make everyone an immediate expert.
This is only possible when you view AI as an architecture, not as a feature.