Gadaa Ask

Wrong answers don’t sound wrong.

Ask once. Several AI models answer separately — and you see exactly where they disagree.

Put to the council

When a key employee gives notice, should you counteroffer?

  • Nemotron-3 UltraNVIDIA
  • GLM-5.2Zhipu AI
  • Kimi K2.6Moonshot AI
  • Mistral Large 3Mistral AI
  • Qwen3.5 397BAlibaba

Diversity anchor — trained on independent data, fails differently

The answer is almost always no — and the reasoning behind that 'almost' is worth understanding precisely, because the exceptions are narrower than most managers instinctively believe.

Why counteroffers fail

By the time a key employee hands in their notice, they have typically been mentally disengaged for months. The resignation is the public announcement of a private decision that was already made. A counteroffer — usually a salary increase, a title change, or a promise of future opportunity — addresses the stated reason for leaving, not the actual one. The actual reasons are almost always relational, cultural, or about trajectory: a manager they don’t trust, a ceiling they can see, a sense that their…

High confidenceStrong agreement, open questions remain
Deliberated by 5 AI models with diverse training lineagesFull record →

Confidence tiers

Every answer carries one of four marks. Read the mark before the answer — the weight of the stamp tells you how much agreement was behind it.

Strong consensus

Every model agreed, including ones built by different labs on different data. That's rare. It still isn't proof.

High confidence

They agreed, but not in a way that rules out a shared blind spot.

Advisory

They didn't converge. You're getting the range of positions, not an answer.

Dissent

Models argued opposite sides of the same question. That's the finding, not a failure.

When they disagree

A single confident answer tells you what to think — but not where the evidence runs out.

When several models disagree, the split itself is the signal: each side holds part of the picture, and the crux is the one question only you can answer.

Gadaa doesn’t smooth over the disagreement. It shows you exactly where they diverge — so the judgment call stays yours.

Put to the council

For a newly created senior function, is it better to hire externally or promote internally and backfill?

  • Nemotron-3 UltraNVIDIA
  • GLM-5.2Zhipu AI
  • Kimi K2.6Moonshot AI
  • Mistral Large 3Mistral AI
  • Qwen3.5 397BAlibaba

Diversity anchor — trained on independent data, fails differently

The honest answer is: it depends on one pivotal variable, and the inputs split cleanly along it.

The crux

The disagreement is not about which factors matter — everyone agrees on the same list: candidate quality, domain expertise, cultural fit, speed to productivity, cost, retention signaling, and political capital. The disagreement is about which factor is structurally decisive when you're standing up a net-new function.

Five independent assessments split into three clusters: two favor external hiring as the default for a newly created senior function, two conditionally favor internal promotion with backfill given a credible candidate, and one conditionally favors internal promotion on grounds of speed, retention, and cost. No position commands a clear majority, and the two 'conditional' clusters share a premise — internal promotion is better if a strong candidate exists — while differing on how much weight to give that condition versus the structural argument for external expertise.

AdvisoryQualified confidenceValues dispute
Deliberated by 5 AI models with diverse training lineagesFull record →

Who this is for

This is for you if

  • You’ve defended a recommendation that turned out to be wrong. It sounded right when you made it.
  • You need the counterargument before the room does.
  • A question you know is contested is more useful to you than a clean answer that hides the contest.

This isn’t for you if

  • You want the answer, not the argument.
  • You need facts retrieved, not judgment weighed.
  • You want a tool that decides for you. This one won’t.

How it works

You ask once.

One question, the way you’d ask a colleague.

Several models answer separately.

They work independently and never see each other’s responses. Nothing is averaged and nothing is smoothed over.

You get one answer, and the receipts.

A single readable response, plus a mark showing how much agreement was actually behind it — and where it broke down.

The models come from different labs, trained on different data, and they fail in different ways. That’s the point: when they agree, it means something. When they don’t, you should know.

Pricing

TrialFreeCore$24.99/moPlus$59.99/mo
Credits50 to start100 a month200 a month
CouncilOpenOpenOpen or frontier —you choose, question by question

An open-council question costs 1 credit. A frontier-council question costs 3. You decide which questions are worth the frontier.

Core is for asking. Plus is for deciding which questions deserve the more expensive room.

What happens when I run out?

Nothing disappears. You keep everything you’ve run, and unused credits roll over. The council stops convening until your next month begins.

No card required.

Where it comes from

Gadaa is an Oromo system for arriving at truth that holds.

No single source sees all of it. An event, a position, a decision — there is always more than one angle, and the truth is more whole when more of them are accounted for. So the answer is reached in council, not handed down by one voice.

It carries a principle called Safuu: you meet every question the same way, with the same moral seriousness, and you stay accountable for the answer you land on.

That is why this is a council and not a model. But it asks something of you, too. The point is not to find the voice that already agrees with you — it is to be willing to hear the one that doesn’t.