01 / CONTINUITY
Identity that carries forward
Keeping an AI colleague's role, working history and relationships consistent even when the underlying model, session or device changes.
Research lab · In development
Persistent AI teams,
under human authority.
리안그룹 연구소 (RIAN Group Research Lab) is developing a company operating system for persistent AI teams: AI colleagues that keep their identity across models, sessions and devices, learn from real work, and collaborate with bounded judgment and evidence-based review.
Early-stage and pre-revenue. This is a research and development effort, not a released product or service.
RESEARCH THEMESCONCEPT
01 / CONTINUITYIdentity across models & sessions
02 / EXPERIENCEWork-derived memory
03 / AGENCYBounded judgment
04 / ASSURANCEEvidence-based review
Roles · Permissions · EvidenceFinal authority stays with people
Concept illustration only. Not live data or shipped capabilities.
What we studyContinuity that outlives a single session.Growth earned from real, reviewed work.
About the lab
An early-stage lab working on how AI teams can be trusted over time.
AI collaboration is often organized around individual conversations. We are exploring what it takes for AI team members to behave more like long-term colleagues: keeping a stable identity, carrying forward what they learned on the job, and working within clear roles.
We treat human authority as a design requirement, not an afterthought. AI contributors can offer independent perspectives and even disagree, but decisions with real consequences remain with accountable people.
- StageResearch & development
- Commercial statusPre-revenue
- Public productNot available
Research direction
A team, not a stack of prompts.
Our work focuses on durable, accountable collaboration rather than personas that only appear persistent.
02 / LEARNING
Memory from real work
Turning experience from actual tasks into memory and, after verification, into shared organizational knowledge, instead of unreviewed self-modification.
03 / ACCOUNTABILITY
Bounded judgment, reviewed by evidence
Designing roles, permissions and review practices so AI team members can exercise judgment and check each other's work while people keep final authority.
Development-stage projects
What we are building now
All projects below are in development. None is offered as a public service, and there is no public access, pricing or API.
Company operating system for AI teams
The central effort of the lab: a foundation for organizing persistent AI team members with roles, permissions, shared practices and human oversight.
Identity continuity
Methods for keeping an AI colleague recognizably the same across different models, sessions and devices.
Work-derived memory & shared learning
Ways to capture lessons from real work, and to promote only verified lessons into capabilities the whole team can use.
Evidence-based review
Review practices in which AI team members independently check work against evidence, with outcomes surfaced to accountable people.
Descriptions reflect current research goals and may change as the work evolves.
Contact
Interested in our research?
We welcome notes from researchers, builders and anyone curious about persistent, accountable AI teams. We are not taking orders, signups or customer onboarding at this stage.