An AI council for consequential questions
Konsil
Hard questions deserve more than one answer.
Ask once. Independent AI advisors answer separately, challenge each other's assumptions, and produce one decision memo: the recommendation, the dissent, the real tradeoffs, and what would change the answer.
Why it works
A second opinion is useful. A structured council is better.
A single model can sound certain while missing the frame. Konsil is designed for questions where the answer depends on judgment: risk, values, timing, hidden assumptions, and what you are probably underweighting.
Question
Should I pay off my mortgage early or invest the extra cash instead?
Invest if your mortgage rate is under ~5%; pay it down if it's above ~7%.
The math hinges on the spread between your after-tax mortgage rate and your expected after-tax investment return. Under ~5%, long-run diversified equity returns (~7–10%) reliably beat the guaranteed return of prepayment, while above ~7% the risk-free return from killing the debt typically wins.
- Behavioral risk can override the math: if the extra cash gets spent rather than invested, the guaranteed debt paydown beats a theoretical equity return.
- Investing accepts market volatility and sequence-of-returns risk in exchange for higher expected long-run wealth.
- Choosing equities over prepayment forgoes the psychological relief of a smaller fixed obligation.
- Prepayment converts liquid, accessible dollars into home equity that's hard to tap without selling or borrowing.
- Whether mortgage interest remains deductible for you, since taking the standard deduction raises your effective rate.
- Your time horizon: as retirement nears, the case for eliminating the payment strengthens.
- If risk-free yields (Treasuries, HYSA) rise above your mortgage rate, prepayment loses its certainty advantage.
- If you're within ~5–10 years of retirement on a fixed income, eliminating the payment becomes more valuable.
Why not just ChatGPT?
Thinking longer is not the same as hearing another perspective.
A single model can critique itself while staying inside the same frame: the same assumptions, the same priors, the same blind spots. Konsil takes independent first takes from different model families before any advisor sees the others, then forces the disagreement into the open before the synthesis is written.
Inside one model
- Self-critique using the same assumptions
- One lab's training philosophy, one set of blind spots
- Confidence without contest
Across the council
- Independent first takes, before they see each other
- Different model families, different blind spots
- Disagreement made visible, then weighed
Best fit
Use Konsil when the obvious answer feels too thin.
Take the startup offer or keep the stable role?
Pay off the mortgage or invest the difference?
Move closer to aging parents, or stay where the work is?
Take a sabbatical or push through the burnout?
Private school, or save the tuition?
Cut off a toxic family member, or keep the peace?
Career and money: decisions where the spreadsheet is necessary but not sufficient.
Strategy and work: plans that need a skeptic before they need execution.
Life tradeoffs: choices where values, risk, and second-order effects collide.
The name
In German medicine, a Konsil convenes specialists from different disciplines to deliberate on a tough case and settle on a diagnosis.
Same shape, applied to whatever question you bring: a panel of experts, structured disagreement, a verdict you can stand behind.
Max
Founder, Konsil
From the founder
When I had a hard question that really mattered, I'd ask it to GPT, Claude, and Gemini in parallel, then tell each one what “advisors A and B” had said and ask if they wanted to revise.
Two or three rounds in, the answers would converge on something I could actually trust. What started as an experiment to automate that workflow became Konsil.
Bring a hard question. Leave with a clearer one.
Sign in with Google to convene your council. No waitlist, no approval queue: bring your question and start.