Esothel
terminal · unreleased
See how language shapes thought.
- Instrument
- M.I.N.D.
- Terminal
- standing by
- Research
- continuous
- Access
- by invitation · public to come
- Stage
- pre-seed · building
A private applied-research laboratory. We build defensive systems that detect manipulation in language: not after the damage is done, but while it is happening.
We build the whole stack ourselves, from the formalism up: a language for describing manipulation, the engine that measures it, and the suite a team needs to work with what it finds. Built for those who defend people from influence: platforms, newsrooms, institutions, defense.
- HPL · APL
- The formalism. Two languages that describe persuasion as triggers, states and loops. SDKs in Python and JavaScript, one conformance suite.
- M.I.N.D.
- The engine. Layered modules score text against a published reference, with the misses published beside the hits. A measurement, never a verdict.
- Terminal
- The console. Plain words in; a person approves every change with a one-time code; every step leaves a receipt on a chain that verifies.
- Interfaces
- APIs, connectors and an MCP server. Local models, provider models or any agent can observe and propose. Approval stays in a human seat.
- Research
- Campaign tracing, case records, standing literature assignments, and the neural side of how language is read. All of it on public text.
- Scope
- Public information only.
Why
Language that reaches you has usually been structured by someone, and increasingly by models, to shape attention, emotion or a decision. We build instruments that measure that structure as it occurs: layered detection of engineered pressure, a score from 0 to 1, and a trace of the features that raised it. The goal is not to restore automatic trust. It is to make the pressure visible while the reader still decides. A clearer channel for reading, not a proxy for deciding.
Access
Unreleased. Public access, in tiers, follows calibration; parts of the engine will be open-sourced after that.