We tell you which model wrote your answer

Most AI products hide their machinery. The answer arrives as if from a neutral, authoritative place — no name on the door, no indication of who thought through the question or what that thinking was built on. The interface is polished precisely to make you stop asking where things come from.

We made a different choice. We tell you which model produced your answer. And we think that choice matters more than it might appear.

Why provenance is not a technical detail

When a colleague answers a question, you automatically factor in what you know about them. Their background, their tendencies, their particular strengths and blind spots. You know whether to take their number at face value or push back. You know when to get a second opinion. This is not cynicism — it is the ordinary calibration of trust that makes collaboration work.

The same calibration applies to AI models. Different models have different strengths, different knowledge cutoffs, different tendencies when the evidence is thin. A model that is brilliant at synthesis may be weaker at precise factual recall. A model tuned for caution may hedge where directness would serve you better. These are real differences, and they are relevant to whether you should act on what you read.

When the model is hidden, you cannot calibrate. You have to treat every answer as if it came from the same source with the same reliability. That is not a neutral choice — it is a choice that asks you to extend more trust than the situation warrants.

What we show, and why

In askFinz, when you receive an answer, you can see which model produced it. You can see the sources the answer drew on. You can see when a claim comes from retrieved material versus from the model's own reasoning. On models, you can read directly about the capabilities and known limitations of the models available.

We do this because we think visibility is the honest baseline, not a premium feature. The alternative — presenting AI output as simply "the answer," attribution-free — asks users to outsource their critical judgement to us. We would rather help you keep it.

There is also a practical reason. When you know which model answered you, you can make a meaningful choice about what to do with the answer. You can decide whether to use a different model for a second pass. You can decide how much verification the answer needs before you act on it. You can build an accurate sense of where different models serve you well over time. That kind of learning is only possible if the information is in front of you.

Citations are not a footnote

The same logic applies to sources. A research result that cites its sources is not just more rigorous — it is more useful. You can follow the thread. You can check whether the source actually says what the summary claims. You can update your view if a source turns out to be less authoritative than it appeared. A result without sources gives you a conclusion without the means to test it.

We have found, consistently, that users who can see the sources engage with results differently. They catch errors. They find angles the summary missed. They come away with a view that is genuinely theirs rather than a view they absorbed without examination.

This is what we mean when we talk about trust at askFinz. Not a privacy checkbox or a security certificate. The specific, concrete act of giving you the information you need to make up your own mind.

The honest answer is a traceable one

There is a version of AI that is confident without being accountable. It sounds authoritative. It does not leave a trail. It is very good at making you feel like the question has been answered.

We are trying to build the other version. The one that tells you where it looked, who wrote the output, and what it is uncertain about. The one that treats your ability to check its work as a feature, not an obstacle.

That version is a little less magical. We think it is a great deal more useful.