Many ideas, judged fairly. The best ones, shortlisted by people.
Give Nexus a problem and it generates ideas in several different ways, then has separate models compare them and argue against them. Ideas are checked against prior art before anyone gets attached to them, and people approve the shortlist.


Ideas compared by one model, challenged by another.
Choose which models compare ideas and which object to them. Keeping them apart stops one model’s habits from deciding what counts as a good idea.
- Judges and critics chosen separately
- Different model families where it matters
- Scores with reasons


Checked against what exists, before anyone gets attached.
Ideas are checked against the outside sources you allow. You see exactly what leaves the company for each search.
- Sources chosen by you
- What leaves the company, shown
- Matches linked to each idea


People choose the shortlist. Budgets keep runs honest.
Decide how many people must approve a shortlist before it is shared, and set limits per run and per team.
Shortlist approvals
- Approvers set per team
- Shared only when approved
- Each decision recorded


Budgets for idea runs
- Limits per run
- Team budgets
- Runs stop before they overspend


About ideas.
What kind of problems is it for?
New products and services, process changes, ways round a constraint, names, offers. Anything where more good options help.
Does it send our problem outside?
Only to the sources and models you allow, and you can see what leaves for each run.
Go deeper
Bring a problem worth solving.
We will run it through Ideas with you and show how the shortlist was reached.