Bizarus.

About Bizarus AI

Our biggest obstacle to finding the truth isn't other people disagreeing with us. It's wanting to believe we're right.

Sketch: a figure seated on a cliff watching the sunrise
Bizarus (noun)
A philosophy of augmented reasoning in which human judgment stays in charge, and the tools exist to help you see further, not to see for you.

The name Bizarus predates this project. Read in the spirit of what the platform does, it fits: good research usually starts with something that seems bizarre, a contradiction, a finding that doesn't fit, an answer that seems too easy. The discipline is refusing to look away from that.

About Bizarus

Every meaningful piece of research starts the same way: someone refuses to accept the first answer. They ask a sharper question, go looking for what's actually true, and hold their own thinking to a standard higher than good enough. That refusal, more than any method or framework, is what research actually is.

Bizarus AI began as an attempt to protect that refusal from erosion. Not to replace the thinking, but to hold the discipline around it steady: the search that shouldn't be skipped, the source that shouldn't be trusted without checking, the contradiction that shouldn't be quietly ignored because the deadline is close. These are small failures, individually. Together, they're how good research quietly becomes mediocre research, one shortcut at a time.

I built this first for my own doctoral work, out of real frustration with how much of that discipline lived in scattered spreadsheets and half remembered rules, easy to hold onto in principle and easy to lose under real pressure. What started as a personal tool for keeping my own thinking honest has grown into something I believe is worth building properly, and sharing the reasoning behind, in public.

I write under the name Bizarus, one I've carried for a long time, well before this project existed. It's simply the name this work is signed with.

Values
01
Augment, never replace. AI gathers, organizes, and challenges. The human decides what counts as evidence and what conclusion it points to. That boundary doesn't move.
02
Rigor over speed. A faster wrong answer isn't progress. Every shortcut protects time for the thinking that matters, not to skip it.
03
Do the simple thing that works. Cheapest robust foundation, simplest working solution. No feature earns its place by being impressive; it earns its place by helping someone think better.
04
Honest about what this is. No promised timeline, no overstated capability, no pretending this is further along than it is.
Project status

Bizarus AI is a living project. It doesn't have a finish line where development stops. It keeps growing, the way real research does.

The core of it, where the actual research work happens, is being used every day, quietly and carefully. It isn't open to others yet. That's not because it's unfinished and waiting for a launch date. It's because getting this right matters more than getting it out.

Some of what's been built along the way is already here for anyone to use. Terminology, Tools, Learn, the Reading Room, and a growing set of templates and guides, all free, no login required. What isn't here yet is the research environment itself. More may open the same way over time, as pieces of it prove themselves ready on their own.

Whether it opens up to other researchers one day, and what that looks like, will be decided thoughtfully, when the time genuinely feels right. There's no promise of a date, because making one up wouldn't be honest.

If you're curious where this goes, Terminology, Research, Blog, and News are the best places to follow along.

This isn't a business, not yet, there's no ad model or paywall behind any of it. If something here has been genuinely useful, you're welcome to support keeping it running.

For anything else, including interest in the project's future direction, feel free to reach out. Prefer to support the project directly? Support this project.

Brand mark
Favicon markFavicon, 32px
App icon markApp icon / avatar