About six months ago, we stopped designing dashboards for our clients.
Now, that would have sounded nuts to me a few years ago, but the world has changed. But our mission has not. The mission was never really about dashboards. It was about enabling better and faster decisions, and dashboards were one way to do that.
So, let me share with you what has changed and, perhaps, how you need to adapt how you work with data.
What has changed
In the past, the best we could hope for was to equip people with the data they needed to make decisions and, ultimately, to act. However, it was very hard to know if people did act, other than through proxies. Even then, it required expertise in change management to drive adoption, and that would often fall to the data teams to be those change champions, which many of them were never really equipped to be.
We are now at a place where those actions can be measured, accelerated, and automated. Dashboard adoption is no longer the primary success metric because now we can measure Decision Velocity, the speed at which decisions are being made and acted upon. That is a much closer way to measure real value.
Decision Velocity
This is really the heart of the shift. For years, analytics teams were judged by activity. How many dashboards shipped. How many reports got built. How many models were deployed. But those things only tell you that work was done. They do not tell you whether the business actually moved faster or made better decisions.
Decision Velocity is a much better measure. Put simply, it is this: how fast can your business go from “What should we do here?” to “Here is what we are going to do,” and then actually do it?
That question matters much more than whether someone opened a dashboard. And by “meaningful business question,” I do not mean a curiosity question. I mean the kinds of questions that actually matter to a business. Questions like:
- Why is churn going up, and what should we do about it?
- Which locations are underperforming, and what action should we take?
- Which claims should be flagged before payout?
- Where are we losing time, margin, or customers, and what can we do right now?
Decision Velocity is really about how quickly those kinds of questions move through the full loop. The question gets asked. The data gets analyzed. A decision gets made. Action gets taken. Then the outcome gets reviewed so the next decision can be even better.
That is the loop. The best teams are not the ones producing the most artifacts. They are the ones helping the business move through that loop faster, more safely, and with less friction. That is the bigger shift. It is not about more analytics output. It is about better decision support.
And this is why I believe Decision Velocity is the better measure now. It exposes the real bottlenecks. If decisions are taking too long, you can no longer hide behind the fact that lots of dashboards were published or that a backlog was cleared. You have to ask where the slowdown is. Is it taking too long to frame the question? Too long to get to the data? Too long to explain the answer clearly enough for someone to act? Too long to move from recommendation to execution?
That is a much more honest way to look at performance.
Where analytics goes from here
Instead of building dashboards, we build decision-intelligence apps. They integrate with all the existing systems, pull in all the data from those systems, address any gaps in process and workflow, and deliver a tailored interface to the target end user. Add in a suite of governed AI agents and your data now has a path to action.
The point is, analytics can now connect to workflow, process, APIs, approvals, and action. I’m not talking about using ChatGPT or Copilot to help summarize things or do research. That was cool about two years ago. Things are completely different now.
We can now build governed agentic systems that connect data, workflow, approvals, and action. Those systems can run in controlled cloud environments or on-prem infrastructure, depending on the sensitivity of the work.
What became obvious once we started building
In 2025, we invested in buying our own servers and GPUs. This was a significant decision and a bit of a gamble at the time, but it is one that has paid back dividends. Owning our own infrastructure allowed us to experiment relentlessly. We failed repeatedly, time and again, but we could do that because we didn’t have massive frontier model costs looming overhead. We could do it all under the CAPEX spend we had made, 24/7.
We learned by building, not by spectating. We found out what actually works versus what is just a vanity project, especially in the past few months with the advent of OpenClaw, as well as Nvidia’s entry into the agentic framework space along with several of the other big players.
What this looks like in practice
To give you a sense of how much change has happened, this time last year I was running a dashboard design workshop, delivering data viz training and supporting multiple clients on their dashboard maintenance needs.
It is completely different now. As I look at our projects, here is a snapshot of what we are building:
Insurance Verification in a Dental Practice
If you ever get to talk to an administrator at a dental practice, they will tell you one of the biggest challenges they have is verifying their patients’ insurance coverage. Indeed, it can occupy about half of the time for the front desk staff. It is incredibly manual, tedious, often involves faxes (yes, I know!), and it is time intensive. No one enjoys this work, and it takes the team away from personal engagement with their patients.
We are building a local AI agent that handles all of this process for them. Why local, you might ask, and not on the cloud?
Well, you could do it on the cloud with one that is appropriately HIPAA-compliant, but we have learned the hard way that local is often preferable, especially in terms of security and sensitivity of data. But there is more to it than that. We found that abstracting everything to the cloud makes it harder for people to relate to what is going on. There is a change management benefit to almost “humanizing” the AI agent and making it appear physically present in the office, doing work that no one likes doing.
The part most companies will get wrong
AI agents can go completely rogue. We’ve had our fair share of learnings, not quite to the degree of the public cautionary tales now making the rounds, but it has certainly been a journey. It is why we developed a whole framework around training agents before they are allowed to be deployed to our clients.
The agents go through over 30 trials, tests, and learnings. They can even evolve, share knowledge between each other where appropriate, and produce all the necessary data for complete tracking. Even then, in production, we still have guardrails to restrict the potential damage an agent can do. It really is a new world.
How data teams need to evolve
While we don’t do dashboard design workshops much anymore, we are doing workshops to enable data teams to leverage AI in the same way we are using it, that is, to rapidly build apps that unify the user experience across multiple tools and data sources into one place.
I do see this as one of the main ways data teams will be evolving.
In this workshop, we show how each of the typical data roles evolves in the AI era, and it is hands-on, so your team would be getting to try out workflows in the session, not just theory.
And this is one of the most important things I have come to believe: roles are not really what transform, capabilities are. The analyst, BI developer, engineer, data scientist, and PM all still matter, but “good” looks different now. The best teams are the ones that remove friction, connect work to real decisions, and help the business move faster without losing control.
What this means for you
If you are still on the fence about AI, or are not changing how you work fundamentally, now is the time to move. Next month is too late. I really believe that.
Don’t wait for the tide to determine your future while you watch what everyone else is doing.
The businesses that will win are not the ones with the most dashboards, the most pilots, or the loudest AI talking points. They will be the ones that reduce friction, accelerate decision-making, and connect insight to action in a way that people can actually use.
Dashboards were never the real destination. Better and faster decisions were. And now, for the first time, we can measure something much closer to what actually matters.
Not dashboard adoption.
Decision Velocity.