Reasoning that keeps the disagreement
Savanax is our own multi-perspective reasoning engine. It runs a question through more than 20 reasoning perspectives, each grounded in its own source material, and shows where they agree and where they pull apart.
The problem
Ask a general AI assistant a hard question about policy, care or compliance and you get one smooth paragraph. The disagreement that matters, between what the law requires, what a family needs and what a budget allows, has been averaged away before you see it. For a decision-maker that hides the choice.
Savanax keeps every perspective separate instead of averaging them, an approach drawn from research on how some minds resist top-down compression. We asked whether an AI system could be built the same way.
What we built
Savanax, also called Dialectica, is deployed in the cloud and runs today. Each perspective works from its own source material: the legal view reads legislation and decisions, the lived-experience view reads first-person accounts. They do not see each other's answers until a final step maps the tensions and presents the map to a person.
Behind the legal perspective sits a corpus of Australian disability legislation, guidance and decisions, indexed so an answer can cite its source.
Consequences
Who is better or worse off, and by how much
Care and relationships
Who is affected and who depends on whom
Disability studies
Barriers in the environment, not the person
Power and structure
Whose interests the default answer serves
Law and regulation
Obligations, precedent and liability
Lived experience
What it is like from the inside
A sample of the perspectives; there are more than 20 in all.
The judgement it took
The hard part was deciding what to leave out. It is tempting to add a final step that picks a winner. We left it out on purpose: the value is in the map, and a person makes the call. We also chose to make a missing voice visible in the map rather than silently absent.
An illustrative question: "A provider wants to move a participant's day program from Tuesday to Thursday to save on transport. What should be weighed?" The legal perspective raises consent and plan obligations, the economic perspective the saving, the care perspective the routine that would break, and the map shows which of these truly conflict.
The evidence we can show
A live deployment, the perspective map for synthetic questions like the one above, and the source sections an answer cites. We demonstrate it on synthetic questions, or on your own documents under agreement.
What you can buy from this now
Decision-support and knowledge tools built on the same approach, scoped to your own documents and questions.
Find it under AI and software systems.
Why aXai
aXai is a disability-led AI, software and accessibility company. We build AI systems, software, and concepts and prototypes ready to scale, and we deliver them through AI-supported work: skilled people with disability in real, properly paid roles, working to clear quality standards, with AI removing the barriers and people keeping the accountability.
Every project runs on our own infrastructure, the Decision Bus and Project Layer, so each step is planned, signed off by a person and on the record. That is our operating advantage. Read how it works.
Talk to us about it
Tell us the problem you are trying to solve and we will say plainly what we can do with it, and what is planned next.