Newsletter #31: The Enterprise AI race really started 6 or so months ago. If you aren't sprinting, now is the time to start...

The Enterprise AI race really started about 6 months ago.

ChatGPT was launched on November 30, 2022.

Approaching two years later, it’s clear that Enterprise adoption of Generative AI enabled tools / products / software has evolved dramatically in the last 6 or so months.?

Phase I was the first 12-18 months. Including a broad / diverse spectrum of Enterprise understanding, exploration, pilots / POCs, evaluation etc.

This phase was characterized by C-Suite questions across the spectrum such as:

“I don’t get it, this feels like a science project, how is this going to have a meaningful impact on our business?”

“Over the weekend, my daughter and her college classmates were home for the weekend and showed me ChatGPT. This changes everything…how do we incorporate this into our business?”

“I’m not sure where we are in the hype cycle or how real Generative AI is but maybe we should just wait and see, what do you think?”

“Where are the Enterprise case studies?”

Phase II started about 6 months ago.?

Hindsight is 20 / 20 but given the case studies that are now being published along with the Generative AI native enhancements Enterprise SaaS companies are rolling out and the leading indicators of the scale of Enterprise exploration with the help of leading consultancies, Phase II is clearly underway.

The race has started. It's time to sprint.

Deterministic vs Non-Deterministic Products

Over simplifying it but one way to describe the essence of why this race is so important is to consider the difference between products built before ChatGPT vs after.?

More specifically, the implications of being able to design and build products that plug into and benefit from Compound AI Systems.?

Deterministic vs non-deterministic products.

Before ChatGPT, products were built in a deterministic way rooted in the principle of predictability.?

Given the same inputs, they always produce the same outputs.?

The user experience of a deterministic product is characterized by consistency and reliability. Clear input mechanisms (i.e. buttons, commands etc.), predictable responses / outputs to well-defined inputs with consistent performance across multiple uses.?

The relationship between inputs and outputs is predetermined, fixed and can be precisely mapped.

Deterministic products excel in scenarios requiring high reliability, such as financial systems, critical infrastructure, or any application where consistency and predictability are critical.?

However, they may struggle with tasks requiring adaptability or creativity, which is where non-deterministic, AI-driven approaches can offer compelling advantages.

Post ChatGPT, now we get to build products in a way that includes non-deterministic characteristics.?

Non-deterministic products, powered by AI, can produce different outputs for the same input.?

Unlike traditional deterministic software, these products embrace uncertainty and variability, offering adaptive and often creative solutions to complex problems.?

They excel at tasks requiring natural language processing, pattern recognition, and personalized user experiences. Non-deterministic products have the unique ability to learn and improve over time through automated analysis of usage data and feedback loops.

A characteristic that can’t be over emphasized given the benefits including the magic of compounding…

Managing these products requires a shift in mindset from seeking exact solutions to focusing on overall system behavior and continuous improvement.

Needless to say, the implications of more diverse and less constrained inputs, personalization potential, continuous data collection, cycle of improvement, automated analysis and recommendations / ideas of non-deterministic products is what makes this race so compelling.

Yes, this is kind of a big deal…across not just Engineering but across all roles and functions (link ) ;)

All “ChatBots” are most definitely not the same...think of “Chat” as an AI Interface and the “Bot” part as a configurable brain with a broad spectrum of utility

Going forward, think of a ChatBot within the context of an iceberg floating in the ocean as well as a legendary quote from philosopher king, Naval.

“AI means that computers are learning our language, rather than us having to learn theirs.” - Naval

Thanks to the advances in AI systems, we can effectively talk to machines without having to learn how to code, the native language of machines. Benefits include opening up access to a much larger population to enable them to build products.

By now, we've all experienced the magic of ChatGPT - the shock and awe of interfacing directly with a Large Language Model, which isn't too different than talking with a colleague who has different or much deeper expertise.

Oversimplifying it but think of the “chat” part of “ChatBot” as the part of the iceberg above the water that is the interface that enables you to access the AI system below the water that you can’t see but powers the quality of your “ChatBot” experience.?

The “Bot” part being the intelligent system below the waterline aka the tech stack…

To make this real, check out my man Tumas R. answer the question “Bankers need more than just a ChatBot?” in 91 seconds (link ).?

Tumas is the Co-Founder + CTO of Rogo and they’re building “Wall Street’s First AI Analyst”

If you want to get a feel for how fast Enterprise AI is now moving, Accenture + LangChain provide a whole lot of signal

Back in August, I shared some commentary from Benedict Evans embedded in AI With Alec Newsletter #29 (link ) tied to some very compelling data around the “Accenture $1B run rate of Gen AI pilots / POCs” along with some additional data from 贝恩公司 and Gartner .

Fast forward to the most recent 埃森哲 earnings call on October 3rd and the growth in GenAI pilots / POC revenue speaks for itself (link ):

If that’s the macro data point, the micro data point comes from my man Harrison Chase , Co-Founder + CEO of LangChain which provides developers with tools to build Generative AI native applications.

Harrison’s twitter feed is full of powerful case studies describing how their clients are generating tangible business outcomes aka running the race, really well:

Listening to the interview of Harrison’s colleague, Julia Schottenstein on dbt Labs podcast titled “The current state of the AI ecosystem” (link ), Julia reinforces how quickly things are moving in Phase II.?

An especially valuable, front row seat perspective given that they have approximately “a million developers who now use LangChain”:

Mental model for how Enterprise leaders might want to think about accelerating their Phase II embrace of AI

I had the chance to attend Pendomonium, the annual product conference hosted by the rockstars at Pendo.io last week here in Raleigh, NC.

During the “AI Adoption: It’s All About the Value” presentation by Inbal Budowski-Tal and Johnny Ross , they walked us through the way they think about AI that informs the very impressive suite of Generative AI enabled products and features they’ve built and are now rolling out.

My favorite part of the presentation was when Inbal walked us through the “4 Levels of AI” (link ), providing a framework for how to think about AI that is more approachable, especially for non-technical business leaders trying to make sense of what can feel like chaos.

Within each level, there are different levels of Change vs Trust required of an Enterprise which I think Guy Podjarny framed nicely via his mental model described in an excellent blog post.

“Trust indicates how much one needs to trust the product to ‘get it right’” vs “Change indicates how much you need to change the way you work to use this product.” (link )?

AI-aware leadership is as important if not more important than the AI enabled technology itself

Because of the Enterprise consensus that’s developed around how significant and real the AI advances are, I believe Phase II of this Enterprise race is being run by everyone.?

Generative AI start-ups are receiving more capital, faster and from sources beyond traditional VCs (link ).

Big Tech attempting to seize the opportunity while being transparent and vocal about their belief that this wave represents an existential threat to their business (link ).

Existing SaaS players, post Product Market Fit see an opportunity to harness Generative AI to unlock outsized growth opportunities by moving faster, smarter than their legacy tech competitors who might be understandably hesitant to innovate due to the Innovator's Dilemma.?

The equivalent is also happening across Fortune 1000 businesses where the buy vs build vs something else equation has become more complicated…

At Pendomonium, I was able to meet the legendary Ethan Mollick who has been touching on all of these trends since day 1.

If you read this newsletter often or follow me on Linkedin or Twitter, you know how stoked I was as he’s as good as it gets on AI Twitter and I'm a MASSIVE fan.

In his most recent blog post - “AI in Organizations: Some tactics” - he unpacks the importance of AI-aware leadership to run this race effectively including this excerpt about Mary Erdoes, CEO of JP Morgan Asset + Wealth Management group (link ):

“Companies will need AI-aware leadership. Our organizations are built around the limitations and benefits of human intelligence, the only form we have had available to us."

"Now, we must figure out how to reconfigure processes and organizational structures that have been developed over decades to take into account the weird ‘intelligence’ of AIs.”

“That requires going beyond R&D to consider organizational structures and goals, and what the role of people and machines are in the organization of the future."

"The right way to do this is not yet clear, but should be something companies, and the consultants and academics who advise them, need to start working on now.”

“Third, model positive use. Executives should be obviously using AI and sharing their use cases with the company. Watch Mary Callahan Erdoes , CEO of J.P. Morgan 's Asset and Wealth Management Group, talk about how the firm is prioritizing AI use at the leadership level, and incorporating their AI experiences into their strategic thinking” (link ).?

After my two day Pendomonium experience, my bull case on Pendo strengthened, a lot

To bring it all full circle, lets bring it back to the rockstars at Pendo.

I believe Competing on Customer Experience is one of the last and only remaining competitive advantages so my Pendo bull case 1.0 (link ) before I attended their two-day annual conference (Pendomonium) was strong.?

After Pendomonium, it’s a lot stronger in part because of the investments they’ve made in Artificial Intelligence aka the way they’re running the race along with a bunch of very positive conversations I had with their Enterprise clients attending the conference, rocking the Pendo pink etc…

In case you aren’t familiar with Pendo (link )…

“Pendo, the most comprehensive product experience platform, allows companies to put product at the center of everything they do. We help teams integrate product intelligence into their organizations to confidently innovate at the speed of changing user needs—taking the guesswork out of delivering the best product experiences.”

“Pendo combines powerful software usage analytics with in-app guidance and user feedback capabilities, enabling even non-technical teams to deliver better product experiences to their customers or employees.”

A few relevant highlights from Pendomonium that drive the point home:?

If the average knowledge worker is provided with “230+ applications” by their employer and spends a disproportionate amount of time working inside those applications, imagine the opportunity to join those data sets, mine them for insights and make personalized recommendations at an AI enabled scale to optimize your tech stack (link )?

Pendo it.

If “AI is taking the marginal cost to build software to zero” imagine how much more important than ever it is to leverage AI to “get to know your users at scale” by aggregating and interpreting billions of data points “to personalize the entire user journey, from discovery all the way through to conversion” and to build products that solve meaningful problems not just building for the sake of building (link )??

Pendo it.

If “insight without action is just noise”, imagine being able to have an intelligent system continuously learning about your 360-degree view of your customers’ needs to “take the guesswork out of product planning” to dramatically increase the productivity of product features designed, built, used and ultimately, business outcomes generated (link )??

Pendo it.

Last but certainly not least, Pendo is headquartered here in Raleigh, North Carolina…so you know how much I love seeing them absolutely crush it along with 2,000 attendees!!?

Enjoy the rest of your Sunday.

Talk soon.

-Alec