After building executive dashboards for years, I realized AI changed the question

A few months ago, during a break at an industry conference, I ended up in one of those side conversations that I kept thinking about long after the conference ended. I was talking with several sales leaders about how AI was beginning to reshape the way they worked — from the CRM and business intelligence tools they relied on every day to the broader enterprise applications that supported their sales process. None of them asked for another dashboard. They didn’t want another tab open in the CRM. They wanted to know, in plain language, which accounts needed attention before the next call. I had spent years leading initiatives that built the reporting infrastructure to answer exactly that question — just spread across three or four different screens. That was the moment I realized the question had changed. Sales teams no longer wanted another place to look. They wanted an answer.

The question no dashboard could answer

For most of my career in enterprise business intelligence, my role has gone well beyond translation. I have led initiatives that brought together data from across the business, helped design the enterprise data architecture underneath executive reporting and worked with sales, finance and operations leaders to turn a business question into a report, a KPI or a dashboard living inside one enterprise application or another. The unspoken assumption behind almost every dashboard I helped build was that the user would go find it, open it, read it correctly and act on it, in the middle of an already full day.

During planning sessions over the past couple of years, I started noticing something I had never heard five years earlier. It was not that the dashboards were wrong. Clicking through three separate systems to prepare for one client call had become a tax nobody had time to pay, and the teams I supported started raising it in one-on-ones and quarterly reviews. At first, I read that feedback as an adoption problem, something a better onboarding session or a cleaner interface could fix. I no longer believe that.

I remember sitting with one of our sales leaders while we walked through his workflow before an important client meeting. We opened the CRM, then a separate reporting application, then a pricing tool, then a forecasting dashboard. Halfway through, he looked at me and asked, “Why can’t one system just tell me what I need to know?” I did not have a good answer for him that day. I have been building toward one ever since, and that single question has reframed how I think about every enterprise application my team touches.

From navigating systems to asking questions

I have spent most of my career supporting sales and partner operations, so I saw this shift first inside sales teams. One request I started hearing repeatedly surprised me. Representatives no longer wanted a better report. They wanted a single conversational entry point that could pull opportunity data, check it against pricing or forecasting numbers, pull in something from an HR system if the deal touched staffing and hand back a recommendation instead of a raw export.

Months later, when Gartner published its prediction that 40% of enterprise applications would carry task-specific AI agents by the end of 2026, up from under 5% a year earlier, it did not surprise me. I had already started seeing exactly that inside the sales organizations I support, well before I saw the number attached to it.

What matters here, and what I have watched happen firsthand, is that the underlying systems of record are not disappearing. The CRM still holds the opportunity data. The HCM platform still owns workforce records. What is changing is the layer sitting on top of the business systems people use every day. CIO’s Bill Doerrfeld’s own reporting on agentic AI backs this up, noting that agents are already updating CRM fields automatically from client interactions, with agent-enriched deals moving through pipeline stages meaningfully faster than the rest. I have watched a version of that same pattern play out with the teams I support. The value is not a flashier report. It is closing the gap between having a question and getting an answer people actually trust.

That word, trust, is where my earlier work on data foundations and this shift toward conversational interfaces meet. A layer that sits across a sales technology stack is only as good as the data underneath it, and a system that can now act on a recommendation, not just display one, raises the stakes on getting that foundation right. I have written before about how AI models fail when the data feeding them was never properly governed. A conversational layer spanning multiple business systems does not reduce that risk. It multiplies it, because the system is no longer just reporting a number back to a human who can apply judgment. Increasingly, it is taking the next step itself.

What this shift asks of BI leaders

I do not think the job of enterprise BI leadership disappears in this shift. I think it moves. For years, a meaningful share of my time went into dashboard design and enterprise data architecture, choosing which metric goes where, how a chart should read and which filters a user needs. Some of that work still matters, but a growing share of my attention now goes into questions that used to sit further down my list. Which systems should an AI layer be allowed to query? What happens when two systems disagree about the same customer? Who is accountable when an agent takes an action instead of simply surfacing a report?

I have started telling my own team something I did not fully believe five years ago. The organizations getting real value from this shift are the ones that treated integration and governance as part of the rollout from day one, not something bolted on once adoption took off. That matches what I have seen leading enterprise analytics for a national network of sales and partner relationships. Connecting a CRM, an HCM platform and a forecasting tool through one conversational layer is a real technical challenge, but it is usually solvable. The harder problem is deciding in advance what that layer is allowed to do once it has access to all of it.

I would tell any BI leader watching this shift the same thing I have started telling my own team. Stop measuring success by how many dashboards get built or how many people log into a portal. Start asking whether the people you support are getting trustworthy answers faster than they were a year ago, inside the tools and language they already use. It has changed how I evaluate success. I no longer ask whether we can build another dashboard. I ask whether we’re helping someone make a better decision faster. If the honest answer is no, the fix is probably not a better dashboard. It is rethinking what sits between your people and the applications they have been navigating on their own for too long.

I still believe in the discipline that built my career. Clean data. Clear ownership. Dashboards that earn a leader’s trust before they earn a login. What has changed is where that discipline gets applied.

It used to live inside the report. Increasingly, it lives inside the conversation. Business intelligence stops being something you check periodically and becomes something that responds to you in real time. The organizations that succeed won’t be the ones that build the most dashboards. They’ll be the ones whose people stop asking where to look because the answer already knows.