ORBIT — Onchain Regulatory & Banking Intelligence Technology — is a Binari incubation: a regtech runtime that evaluates policy against onchain activity in real time. It is in production. We built it, and we operate it.
The thesis
Regulated institutions are moving onto public and permissioned chains faster than their compliance tooling. Most of what exists today is retrospective: batch screening, quarterly reports, alerts that arrive after the fact and can't explain themselves. Our thesis was narrower and harder — compliance has to run at the same speed as the ledger, and every decision it makes has to be defensible to a regulator later. Not a dashboard. A runtime.
That framing has consequences. If an alert can't show its reasoning, it's noise. So we made one property non-negotiable from day one: every alert carries an auditable decision trace — the inputs, the policy version, the evaluation path, the outcome. An analyst or an auditor can replay any decision the system has ever made.
The build
ORBIT was built the way we build everything: a small senior team on studio cadence. Weekly demos of running software, not slideware. No product managers translating between people who write code — the people designing the policy engine were the people shipping it.
We ran the engagement on our own operating system. Aura OS, our internal AI delivery platform, handled the agentic pipelines and evaluation harnesses that let a team this small move at this speed. Development and staging ran on our managed on-premise infrastructure, containerized from day one, with CI/CD ready to promote to cloud at launch — the same discipline we apply on every data-intensive build.
Pipeline before policy
The critical architectural call was build order: pipeline before policy.
The temptation in regtech is to start with the rules — encode the sanctions logic, the exposure thresholds, the typologies — and bolt data ingestion on afterward. That order produces systems that are confidently wrong: sophisticated policies evaluated over stale, incomplete, or unattributed data.
We inverted it. First, a hardened ingestion and normalization pipeline over onchain activity — provenance tracked, latency measured, gaps surfaced rather than papered over. Only once the data layer was trustworthy did we build the policy evaluation engine on top of it. The result:
- Policies evaluate over data whose freshness and lineage are known, so a decision trace actually means something.
- New rules and typologies deploy without touching the ingestion layer.
- When a regulator asks "why did this alert fire," the answer includes what the system knew and when it knew it.
The trade-off was real: for the first stretch of the build, there was no policy engine to demo. We demoed the pipeline instead — throughput, correctness, failure handling. Less glamorous. The right call.
In production
ORBIT's runtime is in production today, evaluating policy against live onchain activity. Every alert it emits carries its decision trace. We didn't hand it off — we operate it, which means the people who made the architectural decisions are the people accountable for them at 3 a.m.
Operating our own incubation keeps us honest. Design flaws surface as pages, not post-mortems. That feedback loop is why the system keeps getting sharper. We wrote more about the underlying approach in onchain regulatory intelligence.
What incubation with us looks like
Incubation is not consulting with extra steps. We put our own engineering, infrastructure, and operating discipline behind a product thesis and carry it to production:
- Fixed-fee discovery first. We scope the thesis and the riskiest technical questions before anyone commits to a build.
- Studio cadence. Weekly demos of working software. Full IP assignment from day one.
- Our infrastructure during development. Managed on-prem dev and staging keeps burn low; CI/CD promotes to AWS, GCP, or Azure at launch.
- We stay for operations. If it's worth building, it's worth running.
We'll also tell you when incubation is the wrong instrument. If your problem is solved by an off-the-shelf tool, buying it beats building it — and we'll say so in the discovery sprint.
Have a regulated-market product thesis that needs a runtime, not a report? Talk to us — we reply within one business day.