At Qlik Connect, one question kept coming up in conversations with leaders: “Which AI vendor should we pick?” But I think that’s the wrong starting point. The better question is: What kind of company do you want to become over the next three years?
Right now, most organizations are heading down one of two paths with AI. Some are bolting it onto existing workflows to improve efficiency. Others are rebuilding parts of their business around it entirely by rethinking products, operations, and how work gets done when intelligence is built into the core of the company.
Those are very different strategies. And they require very different partners.
We picked Anthropic as the first LLM to work with at Qlik this year. But the decision had far less to do with benchmark scores than it did with the direction we believe enterprise AI is heading.
Jack Dorsey put it better than I am about to: "I think most of the industry is thinking about AI as like a co-pilot, as something that is augmented onto, rather than how do you just rebuild your whole company with this as the core."
He's onto something. Most companies are ending up on one of two trajectories with AI, even when it doesn't feel like a choice they made on purpose.
Trajectory one: bolt it on
Take the workflows you already have. Add AI. Speed things up. Carry on.
Most companies are doing this, and honestly, it makes sense. It is fast, easy, and produces nice slides for the board meeting. ("Look, we shaved 11% off ticket time to resolution!")
The trouble is that every one of your competitors is doing the same thing. Same models. Same workflows. By Q3 you will all have made the same upgrade. The improvement from Q1 isn't really an advantage. It is more like a tax everyone paid at the same time.
The longer a company bolts AI onto the work it has always done, the more it starts to resemble every other company doing the same. And eventually, they start to look a lot like the solutions the AI labs themselves are offering, just with different, older branding. That is not a great place to end up.
Trajectory two: rebuild
Look at what your company actually does and ask, honestly, which parts of it would look different if you were starting today. Which work could be done by an agent instead of a person? Which new products you could ship if intelligence sat at the core of your offering instead of being a feature on top? How could you leverage the talent you already have if they didn't have to do x?
This is harder. Slower. The first quarter looks worse than the bolt-on version. (Sorry, there will be no clean slide for the June board meeting.)
And still, it is the path that gives you a real chance of looking different from everyone else once things settle. The teams I see winning right now picked this path over a year ago, and they are quietly running circles around their bolt-on competitors who are still rolling out their 7th iteration of their own copilot.
What this means for who you partner with
Once you have picked your trajectory, the partner question gets a lot simpler.
If you are bolting on, almost any major vendor will do. They are mostly selling the same thing.
One important side topic: if a potential vendor's tools only offer read-only capabilities, that is a sign you are dealing with a laggard, and you should probably keep looking. A read-only AI tool is a museum exhibit, not a partner. The bolt-on-friendly vendors tend to focus on the read-only use cases.
If you are rebuilding, you need two specific things from a partner. First, the freedom for AI to do work inside your stack: open tickets, update records, take action. If your AI can only regurgitate things, you are not rebuilding your business around it. You are decorating with it.
Second, trust — and trust in two places. Trust that what runs behind the demo is your actual context and well-governed data, rich enough to answer the questions your business truly asks, not generic training data dressed up to look like yours. Because an AI without trusted, contextual data isn't intelligent, it's just confident. Equally important is the trust that the vendor will tell you what their tech can’t do yet. Most vendors find that very hard to say. The ones who can — and who treat the data layer as seriously as the model itself — are the ones you want in the room. When AI can churn out a convincing demo in an afternoon, you want solutions, not snake oil.
The thing nobody really wants to say out loud
AI is not a temporary technology cycle. It is reshaping how companies operate, build products, and compete.
The risk is not that organizations fail to adopt AI entirely. It is stopping at incremental efficiency gains while competitors rethink how work gets done at a more fundamental level.
The companies that create lasting advantage will likely be the ones using AI to redesign parts of the business, not just accelerate existing processes.
That bar (context, trust, and the freedom to rebuild) is what we asked of our partner. It is also what we hold ourselves to as a vendor at Qlik, and what we want our customers asking of us.
So if there is a question I would leave you with, it is not which AI vendor to pick. It's, “What do you want your company to look like in three years?”
Pick that first. The vendor question gets a lot easier from there.
In this article:
AI