Sales anon,

As we progress towards the end of the year, it's essential to reflect on the state of GenAI adoption for Enterprise use cases. While technical progress (such as the launch of OpenAI o1) is important and it's always interesting to look at the consumer side of things (the big launches of AI features within Windows/MacOS and Android/iOS), at the end of the day what we care about here is enterprise tech sales.

This post will go deeper in the most recent report by Deloitte on the progress on selling and integrating these solutions for Enterprise use cases.

Show me some AI progress!

There are several important points to understand.

With the strong adoption of AI features within consumer technology and with the significant push from the cloud hyperscalers of positioning these products, there is a strong awareness with decision makers on the general idea and benefits. That’s not the same thing as having vision and intention on realising these benefits.
Improved efficiency and productivity remains the “top of mind” pitch but as I’ve stated previously, it’s a limited and shortsighted view. Doing the same thing with less has diminishing returns; doing more with the same stack is good but not outstanding; the really interesting part is achieving exponential growth by doing something different. The ability of tech sellers to only peddle AI features but instead drive strategic innovation in a customer is what will separate President Club winners from the rest.
There seems to be a better understanding today that GenAI and Enterprise ML is not THE PRODUCT, but technology to enhance and drive different outcomes into existing workflows.

In the use case details provided by the participants of this study, we can clearly see that successfully implemented projects essentially improve existing workflows or create new ones that achieve improved outcomes.

What can be perceived as the most bearish insight from the report is that less than a third of all PoCs are actually progressing to production. I don't think this is particularly interesting statistic if we consider the key factors at play:

Even if the GenAI part of an integration is performing as expected, other parts of the implementation might cost too much or aren't operationally ready.
For many of these companies, their technical teams are mostly shooting in the dark without significant guidance from leadership on what they want to achieve. This is where third parties such as vendors can bring a point of view and ideas that help progress a use case from a PoC to production.

Now, it’s important to remember that this report is produced by Deloitte in order to generate business for them. So it’s not surprising that they identify strategic advisory as the best way to de-risk GenAI initiatives and drive outcomes.

This doesn't make the above chart wrong. Particularly when it comes to complex GenAI implementations, the software vendors need to think through on how to position the project from the lenses of strategy, process, talent, and technology. This is when you move from pitching to influencing and from tactical into strategic.

The data platform angle

One of the primary theses of this newsletter is that the majority of dollar value from GenAI implementations will accrue at the bottom of the stack, i.e., cloud hyperscalers, data platforms, observability, and cybersecurity.

This is slowly starting to become obvious to the bigger players in the industry, as the focus is shifting away from "can I do it with model X" to "how do I build a data strategy that would enable me to get outsized returns on these implementations".

This is very obvious when you look at the significant differences in performance between Databricks and Snowflake. Databricks traditionally focused on selling to data science teams that had complex projects, while Snowflake focused on more traditional data warehouse use cases (as well as a strong edge offering). While "in principle" both companies offered a similar AI roadmap, the reality is that Databricks outperformed in quality of execution both on the R&D side but also in actual implementations.

Customers are specifically coming to Databricks because a) that's where their data sits b) because the teams can advise them on how to optimise and scale their data strategy in a way that can be leveraged more effectively for GenAI implementations.

Now, besides the obvious topic of how to handle structured and unstructured data (plus the quality issues around that), there are a number of additional concerns when it comes to data management and GenAI. What do you ingest, how do you ingest it, who can see it, and how it's distributed are important and complex topics.

One of the most obvious examples is that making an "HR questions bot" might sound like an easy project with obvious productivity returns; however, implementing this in a large organisation that operates in multiple countries can be very risky if local regulations, linguistic nuances, and handling of confidential information are not handled properly. This study identified that more than half of the companies didn't push certain projects to production because of data concerns.

Another topic that tech sales reps would like to avoid but is critical in understanding how customers evaluate big ticket projects is the regulatory risk. AI legislation is a significant topic at the executive level in both the US and the EU, particularly since behind the scenes there are significant attempts to influence decision-making in a more favorable way for software companies.

While it's not our job to solve compliance on behalf of the customer (unless your product specifically addresses GenAI compliance, then by all means do so), it's shortsighted not to qualify compliance in a sales cycle or provide direction on topics they should explore.

“67% of organizations we surveyed said they are increasing investments on Generative AI given strong value seen to date.”

— Deloitte Q3 State of GenAI

At the end of the day the only thing that matters is a) are companies taking GenAI seriously and b) are they expanding their budgets to invest in these projects.

The answer as of 2H’24 is a resounding yes.

Get after it, sales anon.

The Tech Sales Newsletter #59: 2H’24 State of GenAI in Enterprise

Show me some AI progress!

The data platform angle

The Tech Sales Newsletter #60: The downfall of Splunk

The Tech Sales Newsletter #58: Samsara - 6 months later