Investment & commercial support
Support the research, product development and business foundations of Memtric.
Memtric — The Epistemic Path for AI
AI governance for human–AI engineering
AI governance gives human–AI work explicit rules for evidence, review and accountable decisions. RDEP applies that discipline to research and engineering on complex projects.
It defines the rules, evidence requirements, review gates and authority boundaries that govern how uncertain ideas are investigated, decisions are justified, and work is authorized and revised as knowledge develops. RDES (Research Driven Epistemic Supervisor) is being designed to apply RDEP’s rules through that human–agent interaction.
RDEPResearch Driven Epistemic PathDocumented · used in our own developmentThe framework: what must be asked, evidenced and decided, and by whom.
RDESResearch Driven Epistemic SupervisorIn development · current phase: definition and designThe supervisory product and human–agent interaction interface being designed to apply RDEP’s rules during work. An MCP server is the current path under investigation.
AI governance already in use before automation.
Follow the three pilot paths ↗RDEP is already being used manually: people and agents work through its questions, forms, reviews and recorded decisions without waiting for an automated platform. The RDES MCP Server, School of Sardine and BeeMorse pilot paths were started this way. These are working applications of the framework that we can examine and improve as the projects progress.
Help shape what comes next.
We are looking for collaborators to develop and test a framework for sound knowledge and informed decisions in human–AI work, and the software to put it into practice.
Tell us about a possible collaboration ↗
RDEP
Research Driven Epistemic Path
02 ↗How it works
Research, review & human authority
03 ↗RDES and AI agent governance
Research Driven Epistemic Supervisor
04 ↗In practice
The project & its origins
Responsible AI work needs an explainable basis.
For individuals and teams using agents to investigate questions, compare options and develop projects whose decisions need an explainable basis.
See how an answer is examined ↗In a matter of seconds, a language model can produce text that is detailed, coherent and persuasive. That text can look like solid evidence. Fluency alone does not make it evidence.
What looks like analysis may be an assumption. What reads like a conclusion may be a guess. When these blend into a fluent passage, even a careful reader can lose the distinction. The more costly and irreversible the decision, the greater the cost of this mistake.
A shared vision
RDEP is already a documented framework that has been used in the project's own development. RDES is being designed to put that framework into practice through supervision, applicable controls and an interaction interface between humans and AI agents.
The shared goal is human–agent collaboration governed by RDEP throughout investigation, decision-making and authorized action. RDES is intended to help apply the framework’s principles and rules in that interaction, so work proceeds on an explicit basis and within defined authority, and the accountable human can understand and direct it. Preserving records and continuity across agents and conversations serves this wider purpose.
The name Memtric evokes the tricks memory can play: the gap between what we believe we know and what the evidence supports. The epistemic path begins by making that gap visible.
AI governance questions
What is AI governance in simple terms?
AI governance sets out how work involving AI is directed, examined and authorized. In RDEP, evidence, assumptions and unknowns stay distinct, and an accountable human retains reserved decisions, including Gate passage and risk acceptance.
Can you give me an example of AI governance?
Before acting on an agent’s recommendation, a team can check the source, identify unsupported assumptions and bring the required decision to the accountable human. RDEP provides questions, records and review Gates for that process; completing a form does not itself pass a Gate.
Are there any AI governance tools?
Tools can support records, review and checks, while people retain the decisions reserved to them. Memtric is defining and designing RDES to apply RDEP through a human–agent interaction interface; it is not presented as a finished automated product.