Research lab
How it works
Dialectical Reasoning is amplification for experts: several lenses debate a hard question, you can see where claims came from, and the synthesis stays human-readable.
01
Lenses
Several viewpoints look at the same hard question.
Instead of one answer from a single model voice, you pick lenses — like risk, strategy, or ops. Each lens reasons in its own lane, so disagreements stay visible rather than getting averaged away.
02
Provenance
Every claim says how it was known.
Claims are marked as stated (given to us), observed (pulled from the material you provided), or inferred (reasoned from what was available). That label is the difference between trusting a note and mistaking a guess for fact.
03
Multi-lens debate
Lenses argue in structured turns — conflicts become edges.
When two lenses pull in different directions, the workbench records the tension instead of hiding it. You can inspect the clash, not just the winner.
04
Human-readable synthesis
A plain-language reconciliation, not a black-box verdict.
After the debate, you get a readable synthesis: where lenses agreed, where they clashed, and what remains open. It is meant to help an expert decide — not replace the decision.
05
Forkable reasoning
Change an assumption and re-run without losing history.
If a claim rests on a shaky assumption, fork it. The prior branch stays dimmed and inspectable while the new path re-runs under your revised premise.
06
Sparse expert feedback
Mark what matters — a few precise notes beat a flood of clicks.
Experts can rate or annotate individual claims. Sparse feedback steers the next turn without turning the workbench into a survey.
Ready to run a debate?
Sign in with Career Brain, pose a hard question, and watch the lenses work through it.
Start a debate