Last week, a pension administrator told us their team answers the same 12 questions every single day. Pension Dashboards are about to make that problem much bigger.
Around 70 million pension records are expected to be connected to Pension Dashboards by the October 2026 deadline.
The dashboards are almost here. And the Pension Dashboards Programme's own progress report confirms that more than 60 million records had already been connected by December 2025, ahead of the final October 2026 connection deadline.
That is a remarkable technical achievement. But it is also the calm before the storm. The industry has spent years preparing pension data. Far less attention has been given to preparing for what happens once millions of members start asking questions.
This article is for trustees, administrators, and consultants who want to get ahead of what's coming; before October 2026 arrives.
Visibility Creates Questions
For decades, pensions have been largely invisible. Many members didn't know how many pensions they have, where they are, how much they're worth, or even whether they still exist. As of 2024 the Pensions Policy Institute, estimates there are still 3.3 million lost pension pots worth around £31.1bn.
Pension Dashboards change that almost instantly. The moment a member can see multiple pension pots in a single place, pensions stop being abstract. They become real. And when something becomes visible, people start asking questions. Lots of them.
WTW's own research found 71% of scheme managers expect a surge in admin queries post-dashboard.
The technology challenge is almost solved. The member communication challenge is only just beginning.
An early (2026) Financial Times survey found that 80% of working adults would consider using AI for pensions or investment support if it helped them make better financial decisions. Research commissioned by Lloyds Banking Group ,in November 2025, revealed that approximately 28.8 million UK adults, more than half the adult population, are already using AI tools for financial matters.
The questions are coming, and increasingly they are being directed to AI tools that were never designed for regulated pension conversations.
The "Smart Wrong Answer" Problem
This is where the tsunami becomes a governance crisis.
Generic AI tools like ChatGPT deliver responses that are confident, detailed, and often dangerously incorrect, especially in a regulated environment like pensions.
A recent analysis by Aon, in 2026, compared AI outputs with regulated financial advice for real members. The findings highlighted how generic AI can produce responses that are inappropriate in a regulated pensions context:
- Defined benefit transfers: Asked whether to transfer a DB pension, one AI tool recommended proceeding without mentioning Pension Wise guidance or regulated financial advice.
- Pension calculations: AI produced simplified retirement income projections that ignored scheme-specific assumptions and the time value of money.
- False confidence: Rather than acknowledging uncertainty, AI presented speculative answers as if they were authoritative recommendations.
That is the model that works: AI supporting humans, not replacing them.
The Industry Is Already Moving
The wider financial services industry has already recognised this challenge.
In April 2026, Scottish Widows became one of the first major UK providers to pilot an AI-powered investment guidance tool, operating within the FCA's emerging targeted support framework.
For pension schemes and administrators, the message is clear: AI is no longer experimental. It is becoming part of regulated financial services.
The Pensions Regulator (TPR) has also started speaking much more openly about AI. In a 2026 speech, TPR CEO Nausicaa Delfas said:
"AI systems must be governed carefully, with strong oversight, transparency and accountability."
She also stressed that AI should support human judgment, not replace it.
There is also the FCA's PS25/22 targeted support framework and the TPR has signalling it will address AI governance in upcoming guidance.
The direction is clear. The question is shifting from "Should we use AI?" to "How do we use AI responsibly, at scale, and within regulatory requirements?"
What Good Looks Like: A Governed Alternative
The solution is not to ban AI, but to govern its use.
Members already prefer a hybrid model: quick, accurate answers for simple questions, and human support for more complex needs. The challenge for scheme administrators is not whether to use AI, but how to implement it in a compliant, auditable, and pension-specific manner.
This approach is specific. Governed AI for pension schemes is not a generic chatbot added to a language model. It does not provide regulated financial advice or replace trustees, administrators, or specialists. Instead, it manages repeatable, high-volume queries such as transfer value estimates, retirement age questions, and expression of wishes updates, allowing members who need human assistance to receive it more quickly.
Consider a pension scheme with around 8,000 deferred members. An administration team might receive around 600 routine queries each month, many relating to contribution rates, retirement ages, transfer values, or expression of wishes updates. These repetitive questions can consume significant administrator time, delaying more complex cases such as ill-health retirement or pension sharing orders.
With a governed AI assistant embedded in the member portal, many of these routine queries could be resolved instantly, while maintaining a complete audit trail and escalating more complex conversations to administrators when needed. The result is that specialist teams spend less time answering repeat questions and more time supporting members who genuinely need human expertise.
This model is effective: AI addresses routine tasks, while humans focus on areas where their skills are essential.
This is the type of approach MemberBridge was designed to support. It operates within the FCA's targeted support framework (PS25/22), connects with your existing member portal, and can be implemented in three to four weeks without a development project. Every response is governed, logged, and managed by your team.
The Final Question for Trustees and Administrators
The increase in member questions is no longer theoretical. Pension Dashboards will accelerate it. Millions of members will soon see their pensions in one place for the first time. They will ask questions. And increasingly, they will turn to AI for answers, whether you are ready or not.
Will you provide them with a safe, governed path, or will they navigate the risks of generic AI on their own?The October 2026 deadline is closer than it looks.
Pension Dashboards will change how members interact with their pensions. The schemes that prepare for member questions now will be in a much stronger position than those reacting after October 2026.
If you’re preparing for Pension Dashboards and would like to see what governed AI looks like in practice, we’d be happy to arrange a demonstration.
What's your scheme's plan for the member question surge? We'd love to hear what others are doing.



