We are building the system that runs support.
People and AI agents, working the same conversations on one governed system. It is a hard problem and a small team. If that is the kind of work you want to do, we want to hear from you.
Come build it with us
Support is being rebuilt around people and AI agents on one system. We are the ones building it.
BearScope lets a company's team and its AI agents run customer support together: conversations resolved, scored on one rubric, and turned into coaching, whoever handled them. It runs on Fibric, where every AI action is checked before it runs and leaves a receipt you can audit. The interesting problems are everywhere in that sentence, in the product, in the platform under it, and in how a real support team adopts it.
A short list, and we hold to it.
Craft
You care how the thing is built, not just whether it ships. You sweat the empty state, the error state, the keyboard path, the name of the function. You would rather do one part well than three parts roughly. It shows in your work without you having to say it.
Judgment about governed AI
You believe AI is finally good enough to resolve real conversations, and you also believe that capable is not the same as safe. You think clearly about what a model should be allowed to do on its own, what needs a person, and how to prove after the fact what actually happened.
A bias toward real data
You would rather show an honest blank than a confident guess. You check the number before you trust it, you say "I do not know" when you do not, and you are suspicious of a metric that has no provenance. Empty beats wrong is a value you already hold.
Plain words
You can explain a hard system simply, to a teammate or to a customer, without hiding behind jargon. You name the real thing. You are direct when something is broken and direct when it is good. We talk to each other and to our customers the same way.
The way we build the product is the way we run the team.
The principles that make BearScope safe are not just product features. They are how we expect to treat each other and the people who depend on us.
We argue from evidence
Decisions are made on real numbers and real conversations, not on the loudest opinion. If we do not have the data, we say so and go get it, rather than guess and move on.
We are careful with power
We build for the case where something goes wrong, not only the happy path. The default answer to a risky change is no until it is checked. That care is a feature of how we work, not a brake on it.
We respect the human
Software agents do work; people carry judgment and accountability. We do not blur the two, on the team or in the product. The hard calls belong to people, with the context to make them well.
We leave a clear trail
What we did, why, and on whose call should be legible after the fact. We write things down, we review each other's work, and we make it easy to understand what changed and why.
We are not posting a list of seats today. We would still like to meet you.
Rather than invent openings we are not hiring for, here is the honest version: if the work above sounds like yours, send us a note about what you build, what you care about, and where you think you would fit. Real people read it, and we reply.
An honest note: we do not list specific roles, headcount, perks, or office locations here, because we would rather not invent any of it. When we open named positions, this page will say exactly what they are. Until then, the invitation above is real, and so is the reply.