Why Your Customers Should Ask You Their Questions, Not ChatGPT

Learn why your organisation should give the answers, and how EPSO cut email questions with Trusted Answers.

Think about how you looked for information a few years ago. You typed a few keywords, scanned a list of links, clicked, scrolled, and read. Now think about how you do it today. Chances are you ask a question, in full sentences, to an AI tool like ChatGPT, Gemini, or Claude, and you expect an answer.

Customer journeys have changed. People no longer browse; they ask. And that raises a question every organisation should be thinking about: when someone asks a question about your products, your services, or your policies, who should be giving the answer?

Who gives the best answer?

The honest answer is that it should be you.

A general-purpose AI tool knows a little about everything. Your organisation knows exactly what it is about. You know your offering, your rules, your audience, and your content. And in most cases you already have a direct relationship with the people asking, whether they are customers, patients, citizens, or members.

Most organisations already have what they need to give the best possible answer:

  • A website and a digital experience
  • A large body of content
  • Ongoing conversations with the people they serve

The goal, then, is not to optimise your website so the big AI engines can answer better on your behalf. The goal is to provide the best answer yourself.

Not a chatbot. Not a search bar.

When we talk about adding a conversational experience to a website, we are not talking about the chatbot that hides in the corner and implies you are talking to a human. We are also not talking about a search box where you type a keyword and get sent to a page to work it out on your own.

A true conversational experience works differently. The visitor asks a question and receives an answer, built from the content already on your website. Classical search results can support it, pointing the visitor to the right pages and sources. But the work is done for them: the answer is summarized and delivered directly.

The effects are immediate:

  • The user is helped right away: No hunting through menus and PDFs.
  • Efficiency goes up: Without an answer, a visitor might pick up the phone, send an email, or simply leave. A good answer on the spot prevents all three.
  • You build a direct relationship: Every question tells you something about what your audience cares about and needs.

For commercial organisations, that relationship can open up new opportunities. For service organisations, it is a chance to help people better than before.

Why not just let people use ChatGPT?

They will, of course. But when you let a general tool answer on your behalf, you hand over the relationship. You also give up the chance to add context. You know your users better than a general tool does, and the answer they receive from you is fed by you, which means they can trust it.

That is the idea behind Trusted Answers, the first product from Dropsolid AI. It lets organizations craft AI-powered answers, built on their own content, in an open and sovereign way. You choose the model, including open-source ones. It is built directly into your web platform and uses your content. And it comes with observability tools, so you can verify that the context is applied correctly and that the answers are better than what a general platform could provide.

A real result: the EU Personnel Selection Office

The numbers from a real deployment show what this can mean in practice.

The European Union Personnel Selection Office was receiving around 8,000 questions every week. Each one had to be answered by email, and each answer had to be correct, because people were asking about their next job. It was a heavy load for the team.

After deploying Trusted Answers, the number of questions needing human attention dropped by 90%. The team went from handling 8,000 questions to handling just 8% of that volume. The benefits went well beyond saved time:

  • Candidates received answers faster, giving them a better experience.
  • The team was relieved of repetitive work.
  • The genuinely difficult questions received better, more careful answers, because the team could focus on them.

Imagine that kind of result across every organization that answers questions from the public.

Where Dropsolid AI comes from

This product did not appear out of nowhere. Dropsolid started 14 years ago, when websites were reshaping the market. Since then, hundreds of platforms have been built, from small websites to multi-site platforms covering 50 countries.

Two principles have guided the work from the beginning: help organizations connect directly with their end users, because a direct relationship makes everything in the business go better, and do it with open-source building blocks so customers keep their freedom and are never locked in.

Dropsolid AI was spun out to carry those same principles into the AI era. What the big AI providers offer consumers, your organization can offer its own customers.

The return on investment

The business case is straightforward. Productivity goes up because teams spend less time on repetitive questions. Customers get help faster. And a direct relationship with your audience creates room for growth and commercial opportunities that a middleman would capture instead.

There is also a strategic question to consider. If you hand every customer question to a large AI provider, you are effectively making that provider the first point of contact for your customers. Is that really why you built your digital presence and your brand? Most organisations would say no. They want their users to connect directly with them.

Taking the next step

Moving to a conversational model puts things in a different order, and for many organisations it is a significant shift. Working with a partner that has experience in both complex digital experiences and AI makes that shift much easier. If you would like to explore what Trusted Answers could look like for your organisation, get in touch with the Dropsolid AI team.