Gigantic, a big big love
I've spent the last decade building and advising AI startups (tiny companies) that make software for enterprises (ginormous companies). In other words, while I've built products for big companies, my own personal AI adoption has happened mostly at small ones.
Now that I'm working in a big place again, I am thinking a lot about big AI and small AI.
Think big, act small
As a product leader, I understood that enterprises were different from startups, but I mostly internalized those differences like this:
- Wow, it's really hard to get through security review
- Procurement takes forever and everything needs Seven Levels of Approval
- People can only buy software at certain times of year
- Who are these "change management" people? Why are there whole teams for that?
- Etc etc
Like a lot of startup people, I wanted everything to go faster: the consideration and buying process, deployment and rollout, user adoption. But I more or less accepted enterprise constraints as an immovable law of physics.
But even though we thought we understood the constraints, and even though we had worked in enterprise environments ourselves before Textio, we made some mistakes because we fundamentally misunderstood enterprise. We didn't build workflow integrations early enough. We prided ourselves on "instant deployment," not understanding that enterprise IT departments would actually take comfort in things being rolled out in more deliberate stages.
This happened because we were designing for ourselves, not for big companies. We didn't even realize we were doing it. We did it because we had internalized the advantages that small environments have when it comes to AI adoption (and software adoption generally), rather than the advantages that big environments have.
I'm not even talking about the obvious stuff, like bigger companies having bigger budgets. There are two important operational advantages that big companies have when it comes to AI adoption.
Enterprise advantage #1: Scale creates a boatload of data
Big companies are hard to manage. When you have thousands (or tens of thousands or hundreds of thousands) of employees, it's difficult to change things. It can take years to roll out a new tool or process internally.
In the current AI-driven climate where smaller competitors can put AI to work quickly, this seems to place enterprises at a disadvantage. That's sort of true. Except that, once enterprise does get its AI up and running, it can work a whole lot better for them than it does for their competitors at startups.
For instance, imagine a marketing tool that creates new content that is perfectly aligned to your brand guidelines: blogs, slide templates, t-shirts, social posts, and more. Pretty common scenario. Tools like this work by starting with some of your existing content, finding the patterns, and replicating them quickly.
Who do you think gets better output from the tool: a startup that has made three strong customer decks in the last six months, or an enterprise that has made hundreds?
Enterprises create data at scale. Data at scale simply makes AI work better.
Enterprise advantage #2: Scale creates opportunity to measure
Many AI projects fail before they start because the company views AI implementation as a goal unto itself. They measure progress using metrics like "# of employees using AI" and "# of agents built." Savvier companies assess success by measuring the impact that AI has on their core operational metrics. I'm talking about stuff like sales conversion rates, time to ship product, and time to hire.
The truly fantastic thing about enterprises is their scale. For example. it's easier to measure the impact of AI tools (or anything else) on sales conversion rate when you have thousands of opportunities every quarter rather than dozens. It's easier to measure the impact on time to hire if you're hiring 10,000 people a year rather than 10.
Not to mention that, when you're operating at enterprise scale, even small improvements to metrics can be significant for company performance. It can take startups years for AI productivity gains to impact the bottom line; depending on the change, enterprises see impact in months or even weeks or days.
The bottom line: If you're making or rolling out AI products for enterprise, yeah, you''re probably going to need real change management. But if you start by understanding the intrinsic advantages that enterprises bring to the table, what's on the other side of the change can be transformative.
Kieran
If you liked this story, why not subscribe to nerd processor and get the back issues? Also, why not learn to tell data stories of your own?
My latest data stories | Build like a founder | nerdprocessor.com