Persistent heavyweight homes
File-native identity, protocols, knowledge, current state, experience, reflection, and protected integrity records survive the model session driving them.
Aether is a governed AI-agent platform for durable identity, bounded context, selective memory, recoverable execution, and explicit human authority.
The engineering problem
Ordinary sessions lose identity, authority, decisions, and unfinished work at their boundaries. Aether separates those concerns into inspectable systems so an agent can continue without pretending that every old token is still relevant.
File-native identity, protocols, knowledge, current state, experience, reflection, and protected integrity records survive the model session driving them.
The model-facing packet is assembled from provenance-bound atoms with explicit truth, lifecycle, dependencies, omissions, tombstones, and physical budget.
Auto-applied file changes backed by byte-for-byte prior snapshots, deterministic undo/rollback, checkpoints, containment, and bounded recovery.
Operational logs remain evidence. Significant experience and reflection become durable memory through deliberate retention rather than indiscriminate accumulation.
Agents receive explicit authority, context, writable scope, acceptance criteria, evidence requirements, and closure expectations.
Reasoning power never broadens permission. Acceptance, activation, identity ceremonies, and consequential decisions remain attributable and inspectable.
Claims name their evidence layer
As of September 4, 2026, Aether treats limitations as part of the product record. A deterministic fixture pass is valuable evidence, but it is not silently promoted into a production claim.
Planned golden path · ARA-10
The next product milestone joins the proven components into one guided experience. It remains planned until its complete vertical slice passes.
Create a bounded project through a World Model Container.
Select or author a persistent heavyweight identity.
Review the roadmap, staffing, authority, and acceptance gates.
Run bounded work with checkpoints, recovery, and oversight.
Inspect evidence, update continuity, and export the result.
The substrate is not the identity
A model drives a session; the agent's durable home carries its identity, role, authority, memory, and continuity. Model provenance still matters for capability, speed, cost, and evidence.
Master coordination and system-level stewardship under Michel's authority.
Resident local heavyweight for recursive Aether development and long-running work.
Implementation integrity, executable requirements, verification, and continuity.
A manifest-gated identity prepared for future user workflows; its portability experiment is complete, not an internal task to repeat.
Aether Lexicon and research library
Aether now uses a representation spectrum informed by model-family testing. Exact grounding and shared exploratory abstraction benefit from formal structure; readable plaintext carries the operational middle.
Known quantities, identifiers, authority, dates, negation, scope, and acceptance conditions remain explicit and lossless.
Most instructions, explanations, handoffs, and product interactions should remain clear to humans and models.
World models can hold assumptions, disagreements, hypotheses, and accepted unknowns during advanced collaborative inquiry.
The symbolic vocabulary, operators, grammar, and examples.
Open the Lexicon →The project's origins, version history, persistence concepts, and evolving architecture.
Read the origin →Historical symbolic streams, calculators, consensus examples, explainers, and application explorations.
Start with Hello World →Earlier work on continuity, context extension, memory, and durable collaboration.
Read the framework →Reference tables for glyph families and historical plaintext-compression work.
View classifications →A preserved external assessment from an earlier stage of the project.
Read the review →FOUNDER AND HUMAN DIRECTOR
Creator of Aether and final authority for its identities, product direction, acceptance gates, and public release.
Project purpose: not merely to make AI agents run longer, but to let them carry identity, context, memory, authority, and the reasons behind their work across the boundaries where those things are usually lost.