01
Research & synthesis
AI agents can explore constraints, compare sources, structure questions, and turn broad uncertainty into hypotheses and work that can be tested.
About / independent initiative
Romanian Cyber Space Initiative (ROCSI)
Romania / Europe · human direction / AI agents / research / build / verify
Independent research and development rooted in Romania, using AI agents as working infrastructure rather than as a marketing layer.
Applied R&D with AI agents embedded in the way ROCSI works.
AI agents are part of ROCSI's core development loop—from research and prototyping to implementation, testing, review, documentation, and operations. They increase the speed and breadth of execution while human direction, evidence, and accountability stay explicit.
01 / AI agents / operating core
Agents accelerate the loop. People set direction and remain accountable.
ROCSI treats AI agents as part of the development environment itself. They help investigate questions, move between research and implementation, inspect changes, and preserve technical context. The goal is not autonomous output for its own sake. It is a faster and more inspectable path from a problem to evidence, prototypes, and working software.
Agent output must be inspected. It is not authority to be accepted automatically. Human judgement remains responsible for defining the problem, setting scope, making public claims, choosing architecture, handling security-sensitive decisions, evaluating trade-offs, and deciding what is released. ROCSI uses AI agents to extend analytical and execution capacity, not to remove human responsibility from the process.
02
Engineering
They can draft and refactor code, trace dependencies, examine implementations, and accelerate iteration across the stack.
03
Verification
They can review code changes, generate and exercise tests, reproduce faults, and challenge assumptions before changes are accepted.
04
Documentation & operations
They can help keep decisions, release checks, technical documentation, and repeatable operational work aligned with the systems being built.
02 / Why ROCSI exists
Resilience is a systems question before it is a product category.
Digital systems are embedded in social systems. Information integrity changes how people use infrastructure. Interfaces shape institutional behaviour. Technical dependencies become public dependencies. Treating these relationships separately can hide the conditions that make systems fragile.
ROCSI therefore works across civic technology, information integrity, open digital infrastructure, applied intelligence, cyber resilience, and emerging technology while keeping the relationships between those fields visible.
people ↔ society ↔ information ↔ technology ↔ infrastructure
03 / Applied R&D operating model
Research → experiment → build → verify → share.
The process is deliberately evidence-led. We start with uncertainty worth investigating, make assumptions tangible enough to test, carry what survives into engineering, verify the result, and document what was learned. AI agents participate throughout this process and reduce the cost of many iterations, without removing checkpoints or human accountability.
01 / research
Start with a question.
Frame the uncertainty, constraint, or public-interest problem clearly enough to investigate and test.
02 / experiment
Make the experiment tangible.
Turn assumptions into a prototype, interface, data flow, or other artifact that can produce evidence.
03 / build
Learn from the evidence.
Inspect failures, constraints, and trade-offs, then carry what survives into deliberate engineering work.
04 / share
Leave a useful trail.
Share code, notes, decisions, releases, or reproducible evidence where doing so is practical and useful.
04 / How the work is organized
One initiative. Four working surfaces.
ROCSI separates durable questions from active systems and from the technical trail behind them. Collaboration closes the loop by bringing real problems, evidence, code, and deployment context into the work.
01
Research
Durable questions, fields, relationships, and methods that should remain useful beyond a single implementation.
Explore Research →02
Current Work
Active and experimental systems through which broader questions are tested against real constraints.
Explore Current Work →03
Workbench
Engineering decisions, investigations, evidence, observations, and releases that preserve the technical trail.
Open Workbench →04
Collaboration
A route for contributors, organizations, technical teams, research partners, and others who can bring useful problems, data, context, or capabilities into the work.
Explore collaboration →05 / Working principles
Operational habits rather than corporate values.
These principles describe how the work is approached, not claims about institutional perfection.
01
Questions before claims
Frame uncertainty clearly enough to investigate before deciding what the answer or product should be.
02
Evidence before polish
Use prototypes, simulations, data flows, and operational feedback to expose constraints before optimizing presentation.
03
Make boundaries explicit
Keep responsibility, data ownership, external dependencies, and the limits of supported claims visible.
04
Open where openness improves the work
Publish code, decisions, evidence, and releases when doing so increases inspection, contribution, or reuse.
05
Build for inspectability and maintainability
Prefer systems whose state, trade-offs, failure modes, and recovery paths can be understood by the people responsible for them.
06
Claims proportional to evidence
Separate what has been observed or tested from production, safety, security, or impact guarantees that have not been established.
06 / Romania / Europe
Rooted locally. Designed to learn beyond one context.
Romania / Europe
ROCSI starts from Romania because local systems, institutions, infrastructure, and civic conditions provide real constraints rather than abstract use cases. Many of the underlying questions—trust, maintainability, interoperability, resilience, and public usefulness—are broader than one country.
The outlook is therefore Romanian and European at the same time: close enough to practical conditions to test assumptions, open enough for useful lessons to travel.
07 / Grounded in actual work
ROCSI should be understandable through the systems being built.
ROCSI's research areas and Current Work provide the evidence behind the organizational description above. The initiative is easier to understand through the questions it investigates, the experiments it runs, and the systems it is willing to test than through generic capability claims.
Research areas / 06
- Civic Technology
- Trust & Information Integrity
- Open Digital Infrastructure
- AI & Applied Intelligence
- Cyber & Digital Resilience
- Emerging Technologies
Current project records / 03
Platforma Civică
Civic Technology
Connector for NMKR
Open-source integration
Telemetry Attestation Gateway
Trustworthy Infrastructure
08 / Organizational context
An independent Romanian research and development initiative.
ROCSI's public site is organized around the work, the evidence behind it, and the routes through which others can inspect, use, or contribute to what is being built.
- 01Research, Current Work, Workbench, and Collaboration are the primary public surfaces through which ROCSI explains its activity.
- 02Technical material is published when openness improves inspection, contribution, or reuse.
- 03Operational details that should remain private remain private.
09 / Founder / responsibility
Mihai Bărbulescu
ROCSI is currently founder-led and operated directly by Mihai Bărbulescu. He sets the initiative’s applied research and development direction and contributes directly to building the products, systems, and technical infrastructure behind ROCSI’s current work.
He brings more than 20 years of experience across digital products, quality assurance, communications, marketing, and project development, combining product perspective with hands-on technical execution. His current work focuses on applied AI, AI-agent-based systems, civic technology, interoperability, and the practical use of emerging technologies to turn complex problems into tools and systems that can be tested, inspected, and used in real-world contexts.
Through projects such as Platforma Civică and NMKR Connect, this approach translates into working products, integrations, and technical experiments built around concrete constraints, with an emphasis on usefulness, validation, and maintainability.
10 / Collaboration
Bring a real problem, evidence, code, or a concrete constraint.
The most useful next step is often specific: a research question, deployment context, dataset, technical capability, failure case, or problem worth testing in practice. ROCSI is open to collaborations where research, engineering, AI agents, and experimentation can lead to something useful and verifiable.