Continuity
How can a person's thinking remain coherent across tools, models, and time?
Independent research + engineering
Personal Knowledge Infrastructure
Notes, recordings, documents, and AI conversations hold pieces of a person's thinking. The pieces rarely remain coherent as tools change and models disappear.
Deep Well Labs studies the science and infrastructure required to keep the human record durable, attributable, portable, and useful, without surrendering authorship or control.
Our research posture is deliberate: publish the question, preserve the evidence, and keep concepts visibly separate from implemented capability.
How can a person's thinking remain coherent across tools, models, and time?
The original human record must stay distinct from every machine interpretation.
Useful AI should be able to move across commercial, regional, and self-hosted environments.
Provenance, custody, and consent should be visible in the product, not buried in policy.
Underway now
Published as work happens, not after it is finished. Standing and stage are stated plainly so a status update never reads as a finding.
As AI systems keep becoming more capable at reasoning, recommending, and acting, the question that matters is not only what a machine can do, but which architectural choices let a person remain the source of authority over their own record, decisions, interpretations, and delegated actions.
Before a thought can be preserved it has to be captured, and every capture surface loses something: the pause, the hedge, the pressure of the pen, the words said and then unsaid. This study asks which properties of a capture surface let a person record their own thinking with the least loss and the least friction, and which of those properties the existing evidence actually supports.
The public is using AI, marking down the institutions that make it, and beginning to write doctrine about dignity. TEMPER keeps those facts from collapsing into one score: five named instruments that are allowed to disagree, with every number’s parent visible.
Deep Well Labs writes before it builds. These are the documents that argument has produced. Most remain working drafts rather than settled positions. The company treats a document as settled only when it is deliberately promoted.
Why the company exists. Humanity has built infrastructure for communication, computation, and information, but not for the external record of human thought. The appropriate role of technology is not to claim possession of a person's mind, but to help them preserve what they intentionally express, retain its provenance, and govern what machines may infer from it.
The convictions the company was founded on and the ethical commitments that follow from them, including a Christian foundation held as company character rather than as a requirement of anyone who uses the products or a claim of authority over anyone's faith. It commits the company to treating every person as having inherent worth, to never reducing a person to a source of attention, data, or revenue, and to never using faith as decoration for conduct that does not support it.
The rule separating what the company researches from what the product claims. Deep Well Labs asks what could become possible; Origo must answer what can honestly and reliably be delivered now. The charter defines five reality classifications (exists, prototype, simulation, frontier, speculative) and the evidence a research capability must produce before it may enter the product.
A plain-language account of how Origo makes alteration of a record impossible to hide. Every entry is fingerprinted, each fingerprint includes the one before it, and the resulting chain is witnessed outside Origo on a schedule. It is explicit about the limit: tamper-evident, not tamper-proof, with a stated window between one witness record and the next.
The technical companion to the integrity brief, for a reader who intends to check rather than decide whether to trust. Names the export's SHA-256, the dates, the timestamp authority, the exit code, and the conformance corpus a verifier was tested against, so its own claims can be checked.
A survey of what science currently knows, disputes, and does not know about human thought, written so that the company's reasoning rests on the state of the evidence rather than the popular summary of it. Its central finding is stated first: there is no scientific consensus on what a single thought is, which constrains what any system, Origo included, may claim to capture.
The company's founding argument, written to be argued with. It sets out why a person's accumulated thinking needs infrastructure that survives changes of model and vendor, and states the invariants Origo holds itself to. Claims are individually tagged by evidence class (established, inference, hypothesis, implemented, modelled, ambition), so a reader can tell an argument from an achievement.
Available on request : The Continuity ThesisThree rules governing AI authority across every Origo product. AI may not claim ultimate, spiritual, or divine authority. It may not replace human agency, trusted relationships, or qualified professional care. Source records and AI interpretation must remain architecturally distinct. These take precedence over convenience and over product ambition.
Available on request : Human Agency and AI GuardrailsThe themes the company considers worth sustained investigation, and the lifecycle by which a raw idea becomes a question, then evidence, and only sometimes a decision. It is explicit that research does not graduate into implementation automatically: a theme may remain exploratory indefinitely, be narrowed, or be rejected.
Available on request : Research RoadmapResearch, made concrete.
Origo helps people create a durable, user-governed record of their expressed thoughts and observations while keeping original human expression distinguishable from machine-generated interpretation.
Origo is in private alpha and not yet generally available.
Pocket Chronicle, a local-first iPhone application for the Origo Chronicle, is under development for a fall alpha release. It keeps the record encrypted on the device itself, with no account, server, or synchronization.
Origo is a product of Deep Well Labs, Inc.
Visit HelloOrigoIllustration This is a diagram of how Origo separates a person's own words from anything a machine derives from them. It is not a product screenshot and does not depict a real record.
I want somewhere trustworthy for a thought to land before it disappears.
Possible connection: continuity matters first at the moment of capture.
“Origo began with a personal need: somewhere for thoughts to go when the mind would not slow down.”
Deep Well Labs grew from the larger question of how people can preserve and continue their thinking without surrendering ownership, authorship, or control.
This is not about outsourcing thought. It is about giving thought somewhere trustworthy to land.
Trust is not a settings screen. It is the architecture, the interaction, and the visible boundary between a person and every inference made about them.
What a person actually said is the source. Everything derived from it must remain visibly derived.
AI may help find patterns and possibilities. It does not silently become the author of the record.
A person's accumulated knowledge should not become captive to whichever model or vendor is fashionable this quarter.
The system should reduce burden at capture and compound understanding as the record grows.

For researchers, builders, and long-horizon partners.
Deep Well Labs is working at the intersection of cognitive science, knowledge infrastructure, human agency, and AI systems. Serious conversations are welcome.
Start a conversation08 / STAY CONNECTED
Occasional updates on Deep Well Labs research, experiments, and what we're building.