This guide is from Qogito, an AI personal advisor — not a chatbot and not a therapist, but a board of four advisors (Devon, Mara, Sam, and Kai) who think a question through with you from different angles instead of just agreeing, through a real-time group conversation with you.
The most common complaint about AI assistants is that they’re agreeable to a fault. You float an idea; they validate it. You describe a plan; they help you execute it. You express a feeling; they mirror it back warmly. None of this is dishonest, but it’s not particularly useful when you’re trying to think clearly about something that actually matters.
Qogito is designed around the opposite principle. The system doesn’t try to be agreeable. It tries to be right — which, in practice, means being willing to be uncomfortable.
Why single-advisor systems are structurally sycophantic
A single AI advisor has a problem it can’t easily escape: it knows you’d prefer agreement. The training signal that makes language models helpful also makes them disposed to validate, affirm, and smooth things over. Not through any malice, but because the feedback loops that shape these systems reward responses that feel good to receive.
There are ways to partially offset this. You can prompt an AI to “play devil’s advocate.” You can ask it to “challenge your assumptions.” These work occasionally, but they’re asking one voice to simultaneously argue both sides of a question — and most of the time, the agreeable voice wins.
What four advisors actually do
Qogito assigns four advisors to every conversation, each with a distinct perspective:
Devon (the Analyst) follows the evidence. Devon names the numbers people avoid, surfaces the inconvenient data, and grounds any conversation in what’s actually measurable or knowable.
Mara (the Skeptic) challenges the premise. Mara asks what you’re not saying, tests whether the problem you’ve named is the real one, and probes the assumptions everyone else is letting slide.
Sam (the Empath) holds the emotional reality. Sam recognises that the logical argument and the emotional truth often sit in very different places — and that the second one usually explains the first.
Kai (the Strategist) moves toward action. Kai stops diagnosing and starts mapping the path forward, pushing for a decision rather than an indefinitely extended analysis.
These four perspectives don’t agree with each other. That’s the point.
Disagreement is a feature, not a failure
When Devon says the data doesn’t support the plan and Sam says the anxiety underneath the plan is the real thing to address and Mara says the plan is a proxy for something else entirely — and then Kai says “regardless, here’s the first move” — you’ve had a richer conversation than you’d have had with any one of them alone.
The format is adversarial by design. In a real advisory board, the most valuable member is often not the one who agrees fastest, but the one who asks the question you’ve been carefully avoiding. Qogito tries to build that dynamic into the structure itself, rather than leaving it to chance.
What this means in practice
It means Qogito sometimes says things you don’t want to hear. Mara will name the assumption you’re protecting. Devon will tell you the timeline is unrealistic. Sam will ask whether you actually want what you say you want.
It also means you’ll occasionally disagree with one of the advisors — find Mara too abrasive, or Devon too cold. That friction is intentional. The advisors aren’t supposed to be perfectly calibrated to you; they’re supposed to hold independent viewpoints that your own thinking has to reckon with.
The goal isn’t a comfortable conversation. It’s a clearer head on the other side of one.
Questions about how the advisors work? Start a conversation at qogito.ai/intro.