Why Power Pages builders need a different approach now

Here’s something I’ve noticed after years of delivering Power Pages projects: the hardest part usually isn’t the architecture or integrations. It’s everything surrounding them, the web role assignments, permission configurations, cross-environment validation checks, and governance reviews. None of that work is glamorous, and honestly, none of it directly improves the experience for the person actually using the portal.

That’s where agentic AI is starting to change things, and it’s changing them in ways I didn’t fully anticipate when I first started experimenting with it.

While researching this shift, one statistic stood out to me: developers spend up to 30% of their time on setup, rework, and validation instead of solving real business problemsⁱ. I recognized that immediately. On more than a few projects, I’ve watched delivery teams lose days just getting environments to a point where meaningful work could even begin.

And the issue doesn’t stay small for long. One portal is manageable. But three portals across multiple environments, each with different governance requirements and overlapping data models, is where things start to break down.

What I’ve also noticed is that AI now shows up almost everywhere in my workflow. But most AI-assisted work in Power Pages development still stops at drafting content or generating snippets of code. The larger structural decisions, how the site is organized, what data gets exposed, how permissions map to actual user journeys, still fall entirely on delivery teams to manage manually.

That gap is becoming harder to ignore. Compliance expectations are tightening. Security reviewers are asking harder questions earlier in the lifecycle. What once felt like manageable manual configuration now introduces genuine operational risk when handled informally.

Figure 1. Where Power Pages builder time goes — setup, rework, and validation consume up to 30% of delivery effort

What’s changing in Power Pages development right now

Most of the teams I work with are not cutting corners. They’re following the right practices: defining data models early, locking down access rules before front-end work begins, and running security validation before deployment. That hasn’t changed.

What has changed is who or what is helping manage parts of that process, and when those decisions happen.

The difference between standard AI tooling and agentic AI for Power Pages is harder to explain than people expect, so I usually frame it this way: a standard AI tool answers the question you ask. Agentic AI asks the question you didn't think to ask. It connects page structure to Dataverse  tables, permission models to authentication methods, and deployment pipelines to governance policies. Instead of forcing delivery teams to stitch those relationships together manually, it proposes them as a connected system.

For Power Pages teams, that means less dependency on institutional memory and fewer access-model issues surfacing late in delivery.

The builder still owns every decision that reaches production. That part does not change. But they no longer have to reconstruct the entire picture from scratch on every engagement.

Figure 2. How agentic AI connects site structure, Dataverse tables, permission models, web roles, security validation, and deployment into one coherent view

How agentic AI changed the way I build Power Pages sites

Most people assume the biggest benefit is speed. I’d push back on that.

For me, the biggest shift has been confidence. More specifically, it’s the confidence that comes from seeing design decisions presented as a complete system instead of scattered across multiple configuration screens and informal conversations.

When agentic AI proposes a site structure, a data model, and a set of role-based permissions together, I’m reviewing a unified picture instead of assembling one piece by piece. I can immediately see whether the access model genuinely reflects how users will move through the portal.

That visibility changes technical reviews. It changes security conversations. And in regulated environments that matters because nothing should move forward without explicit human review.

What surprised me most was what happened with the time I gained back. I expected to move faster. Instead, I found myself thinking more deeply about whether the data model would still hold up in two years, whether the permission structure could survive organizational change, and whether the portal architecture would remain maintainable for the next team inheriting it.

The quality of the work improved. That wasn't something I originally expected.

Practical recommendations for using agentic AI in Power Pages projects

From what I've seen, agentic AI for Power Pages does not improve every stage of the delivery lifecycle equally. The earlier you introduce it, the more valuable it becomes.

During early design, when teams are debating site structures, mapping permissions, and aligning environments, that’s where I’ve seen the strongest impact. Bringing it in during final validation is usually too late. By then, the decisions it could have improved are already locked in.

There are also areas where AI simply cannot replace human judgment, and I’d be cautious of anyone claiming otherwise.

Interpreting regulatory requirements for a compliance-heavy client still requires human context. Deciding what data should or should not be exposed through a partner portal remains a business decision, not a technical one.

What agentic AI does exceptionally well is reduce friction in the repeatable parts of the process. It handles the assembly work so teams can spend more time applying judgment where it actually matters.

Figure 3. Before and after adopting agentic AI for Power Pages — human accountability remains constant; effort shifts from assembly to governance

Barriers to adoption and a real-world example

The biggest barrier I see is not technical. It’s trust.

I hear the same concern repeatedly: AI introduces risk, reduces oversight, and removes control from the team. I understand why people feel that way. But what I’ve observed in practice is almost the opposite.

The real risk usually lives in undocumented decisions made quickly under delivery pressure. A transparent AI recommendation that can be reviewed, challenged, and audited is often easier to trust than a configuration choice someone made late on a Thursday afternoon because timelines were slipping.

In one of the projects, a mid-size enterprise portal in a compliance-heavy sector, we brought agentic AI in only for non-production, just to test the recommendations before anyone committed to them. The output flagged permission gaps no one on the team had spotted. Not because the team was careless; they were experienced. The gaps were the kind that only become visible when you look at the whole structure at once, rather than one configuration screen at a time.

Every recommendation was reviewed, adjusted where needed, and signed off before it moved up. Architects were more confident in what they were approving. Security reviewers had fewer concerns to raise. And what eventually went live was cleaner, more defensible, and significantly better documented than what we had historically delivered through traditional methods.

AI supported the review process. It did not replace it.

Conclusion: Less assembly, more intent

I'm still accountable for everything that gets delivered. That part hasn't changed.

What's changed is where my time actually goes. Less of it is spent checking boxes and rebuilding configurations I’ve assembled dozens of times before. More of it now goes into the work that is harder to automate: experience design, governance decisions, edge-case validation, and thinking about what a portal should still look like three years from now.

For teams running large-scale Power Pages development programs, that shift is significant. It’s the difference between delivering something that simply works and shipping something that lasts.

And this is where I believe the bigger story begins.

Business Creativity is not just about moving faster with technology. It’s about bringing human insight and intelligent systems together to rethink how work itself gets done. That’s exactly what agentic AI enables in modern portal delivery. It allows teams to Outcreate fragmented configuration management by replacing disconnected manual processes with a more intelligent, connected, and reviewable way of working. Not just improving delivery but reframing how delivery happens in the first place.

The teams that move early on this shift will gain more than speed. They’ll build portals that are more defensible, more maintainable, and better aligned to long-term governance expectations.

That’s what creating new productivity paradigms actually looks like in practice. The shift is already happening. The only question now is whether you’re leading it or catching up to it later.

It’s time to Outcreate.

Reference:

1.        2023 Developer Survey, Stack Overflow: https://survey.stackoverflow.co/2023