Founder / CEO · 8Braid Inc.
Ashley is the founder of 8Braid Inc., creator of 8DB and sole author of its research program. His work focuses on database architecture, post-quantum protection, model-independent information systems, private AI and locally authoritative edge infrastructure.
The database should connect information, intelligence and applications, with security, privacy and owner control built into the same foundation.
Published work
AI With Authority
An AI that can act for you needs more than a spending limit. It needs authority tied to current rules and evidence. That opens useful possibilities, and an uncomfortable question about who controls the controls.
Iterating on Navier-Stokes
We reproduced the formal checks, proved an explicit bound for one construction step, and tested how 8DB handles its evidence. The larger opportunity is to help AI catch checkable errors while it works, before they become the next step's assumptions.
Trust Beyond Ledgers
Could a benefit carry its eligibility rules, spending limits and audit evidence without creating an unnecessary record of a person's life? A proposed application of 8DB explores programmable entitlements, fraud controls and carefully defined disclosure, with the remaining proof and deployment questions in view.
Trustworthy AI Data
Every defence against prompt injection you have read about lives in the prompt, the orchestrator or a filter in front of the data. All of them are software an attacker can talk past. There is one boundary a clever prompt cannot argue with: the agent's process never had the key. 8DB puts clearance and policy inside key derivation, so a request outside policy comes back as ciphertext no matter how persuasive it was. Here is how to build agents on that boundary.
Iterating on Navier-Stokes
The forced Navier-Stokes breakdown proof runs on oscillatory loops with a safety margin. We took one of those loops at its reference profile, found a shape with a margin about twenty percent wider than the best any single sine wave can achieve, then let the research graph choose the next experiment and found a second shape that lowers four of the construction's own cost bounds at the same margin. Every bound is a certified interval, every candidate that lost is kept, and the database did the bookkeeping.
Iterating on Navier-Stokes
In the second week of September 2026 a kernel-checked Lean proof established two of the four Clay alternatives for Navier-Stokes, and a second group published machine-checked forced blow-up for three related equations. Within seventy-two hours we had rebuilt the proofs from source, audited their axioms, matched them to the pinned statements under two kernels, and recorded every step as governed evidence in 8DB with host, time and exit code. This is the map of what is established, what is open, and what a database adds when the frontier of an open problem moves in a week.
Trust Beyond Ledgers
Blockchain made a shared transaction history believable. What if a transaction also carried its authority, policy and supporting evidence? A future factory purchase shows how 8DB could connect those pieces, and which privacy and network guarantees still need to be demonstrated.
Post-Quantum Databases
A database does not become post-quantum because its wire does. The records on disk, the indexes that find them, the backups that restore them and the keys that must hold for decades are the boundary that decides whether your data survives the transition. Here is how to draw that boundary, the questions to put to any vendor, and the measurements showing that the post-quantum cost lands at the boundary and not on every row.
AI With Authority
What if an AI earned its reward by passing an agreed check, including a check that proves your original idea wrong? A thought experiment about commissioning useful, inspectable research.
AI With Authority
A data owner could sell twenty executions of an approved analysis without distributing the underlying dataset. The useful product is a permitted use, with clear limits on what leaves.
One Protected Substrate
A BioTwin research prototype used 8DB's lattice operators to examine eight competing explanations. The useful result was knowing what each possible observation could settle, and what it could not.
One Protected Substrate
8DB includes a cryptographic channel that brings braid structure into key derivation. It is a concrete example of a native data model doing more than storing information.
Iterating on Navier-Stokes
Five months of Navier-Stokes work leaves a lot of documents, and documents are where attractive ideas quietly become premises. We read every accessible file on the subject, 573 of them, and asked one question of each claim: does it survive an exact counterexample? Ten did not. The review also found three real defects in the database itself, fixed them with regression controls, and recovered a formal contribution that older summaries had been calling missing. The ideas survived; the invalid inferences stopped supporting conclusions.
Trustworthy AI Data
Models are scored for accuracy and stores for throughput. The layer that actually decides whether an agent's answer can be acted on, whether its data knows where each fact came from, how sure to be, and when to say it does not know, has no score at all. We are proposing an open, vendor-neutral benchmark for that layer, and we want the people who would score badly on it in the room.
Iterating on Navier-Stokes
The state of an open problem is not a document. It is a graph of statements, each with evidence, each implying others, some still open, and it changes every time someone posts. We built that graph for Navier-Stokes as governed atoms in 8DB: intervals instead of scores, restriction maps that refuse at write time to let a conclusion outrank its premise, typed gaps for what is open, tiers derived from verification events rather than assigned, and a retraction query that answers in one call what falls if a premise is withdrawn. This is the design, with the numbers it holds.
8DB on Apple Silicon
Measured on an M1 with the machine's own power sensors, a photo-ingest workload through 8DB used about 34 joules per gigabyte written against about 130 through SQLite. The reasons are structural, not tuning: a write-optimised tree writes fewer bytes, indexes built at write time remove the background re-indexing daemons, and relevance-gated sync moves only what another device needs. Here is the measurement, the mechanisms, and what they project to for an iPhone's and a Watch's day, with every assumption in the open.
Trustworthy AI Data
Long-running agents are being given memory: a store of what they saw, decided and learned, carried across sessions and users. That store is now the most sensitive database in the building, and it is being built on vector indexes with no notion of who may recall a memory, where it came from, or how to make it gone. 8DB makes recall a cryptographic fact rather than a filter, gives every memory its pedigree, and turns forgetting into something a third party can check.
Trust Beyond Ledgers
Regulators of crypto-asset settlement have stopped accepting screening that runs after the decision and audit logs that summarise it. They want proof that policy was a precondition on the transfer, and the ability to reproduce, byte for byte, why it cleared. That is a data-structure requirement, not a reporting one, and Sovereign Clearinghouse on 8DB is built to satisfy it: every decision a gate, every gate replayable, every supervisor given a lawful view of the same record.
Trust Beyond Ledgers
A deletion request should leave a clear account of what changed and what still depends on it. 8DB's braid work connects a scoped structural removal with evidence about the change.
Evaluation and Assurance
Vendor benchmarks are the least trusted documents in technology, and they deserve it. Ours carry their quantity, their scope and the harness that produced them, and the rule is enforced in code, so a figure that cannot meet it never reaches you. Here is the rule, the incident that made us adopt it, and how to use it against us.
Iterating on Navier-Stokes
A proof assistant tells you a proof matches a statement. It cannot tell you the statement matches the problem. When the Navier-Stokes breakdown proof landed, the statement it was checked against was a Lean encoding of the Clay Institute's prose, written months earlier by someone else. We audited that encoding clause by clause: sixteen faithful, one concern, and three that rested on an argument in English, which we replaced with twenty-two kernel-checked lemmas.
8DB on Apple Silicon
The promise of intelligence that never leaves the phone is only as good as the data layer under it. A model retrieving from a store that cannot say where a fact came from, who may see it, or how sure to be will give private answers that cannot be trusted. 8DB is the retrieval layer built for that setting: multi-anchor search that returned 3.6 times the precision of full-text search on the questions people actually ask, traversal that enforces clearance on every hop, and a privacy registry that decides at write time what may ever leave the device.
Fielded Systems
A data layer for platforms that operate cut off, keep records longer than the algorithms that protect them, and have to be assessed on what can be shown. 8DB puts persistence, governance, protected transfer, recovery and diagnostics in one qualified runtime, leaves no key table on the platform, and answers each of the seven questions an assessor asks with an artifact that already exists. Here is what goes in the binder, question by question.
Iterating on Navier-Stokes
Store a loop from a fluid-dynamics proof as a table of sampled values and you can plot it. Store it as one object with six views, a symbolic formula with exact parameters, a charted manifold, a point cloud in state space and again in derivative space, a weighted field on a declared domain, and an interval certificate, and you can ask which questions the table hides. Here is the loop, view by view, and the query in which two identical states turned out to be moving in different directions.
8DB on Apple Silicon
Every app on an Apple device keeps its data in the same embedded engine, behind a C interface that has not changed in decades. Put 8DB underneath it and the device gets a store that is faster than SQLite on its drop-in write path, carries provenance, confidence and governance as storage columns, answers graph, search, spatial and face-match queries from one substrate, and runs at the same speed on an M1 as on a desktop chip with six times the cache. Here are the measurements, with the host and date on every one.
One Protected Substrate
The same participants and final state can describe very different histories. Native braids give ordered interactions a structure of their own in 8DB.
Iterating on Navier-Stokes
The Lean certificate for forced Boussinesq blow-up contains about twelve hundred machine-generated interval-arithmetic modules that the kernel decides one by one, and its authors budget 100 to 150 gigabytes of memory to replay it. We had 62. The number turned out to be parallelism times per-module footprint, not a requirement of the check, so we ran the build inside a hard memory clamp and let it fail and resume until it fit. Every run left behind a record a stranger can inspect.
Trustworthy AI Data
When an AI system makes a consequential call, someone will ask why. Today the answer is assembled after the fact from logs, traces and the memory of whoever was on shift. In 8DB the answer is a query: the source records, their confidence, the governance state at the time, and a proof a third party can check without trusting anyone who ran the system. A two-week reconstruction becomes a session.
One Protected Substrate
The modern data platform is a relational store, a document store, a vector index, a graph database, an object store and a search engine, stitched together by pipelines and secured by six access models. Every seam is a place where governance, provenance and protection quietly disagree, and every audit finding about a leaked record eventually traces to one. Here is the case for one substrate, written for the sceptic, with the acceptance criteria any consolidation claim should have to meet and what 8DB has demonstrated against them.
Post-Quantum Databases
A quantum computer breaks the envelope around your records, not the bytes inside it. Here is what breaks, what does not, where the real work is, and the three questions that end the pitch.
Post-Quantum Databases
NIST has published the dates: quantum-vulnerable public-key algorithms are deprecated in 2030 and disallowed in 2035. That makes the post-quantum transition the first algorithm migration in history with a known successor already on the calendar. Organisations treating it as a one-time project will run it twice. Here is what it takes to run it once and then flip a switch.
Trustworthy AI Data
Your model did not hallucinate. Your database did. Retrieval taught the industry that an answer can only be as honest as its store, and most stores cannot say how sure they are, where a fact came from, or the difference between unknown and absent. 8DB carries confidence and provenance on every fact, distinguishes eight kinds of nothing, and refuses to answer past its evidence, which changes what the model on top of it is allowed to say.
Trustworthy AI Data
Sanctions screening asks one question of one name at a time. Load the entire public sanctions corpus into a store that treats it as a structure, and different questions become askable: which unlisted entity holds the network together, which subsidiaries of a designated parent were never listed, which nominee held fourteen assets for sixteen months and moved them all in one day. Here is what that looked like on 4.28 million public records, and why every finding arrived with a proof.
8DB on Apple Silicon
8DB gives an application three ways to write. The drop-in path speaks SQLite's C interface, preserves read-your-own-writes, and inserted about 2.9 times faster than SQLite on the same M1 in the same run. The native path inserted in 99 nanoseconds, about fifteen times faster. The third path costs more per call because it computes confidence, provenance, governance, a perceptual hash and cross-modal indexes on every write, replacing a pipeline of separate passes with one. Here is the benchmark and the trade each path makes.
Post-Quantum Databases
The strongest incumbent answer to 'change the algorithm on stored data without an outage' is Oracle's online tablespace rekey and conversion, and it is the comparison that matters. We are measuring 8DB's record-level migration against it on the same records, under a steady live workload, on hardware you can name. The method, the fixtures, the disclosure rules and the pass criteria are here, fixed before a single number is, so that nobody has to trust the winner's account of the race.
One Protected Substrate
In spring 2026 we rendered a full city extract of OpenStreetMap, more than thirteen million entities, on an iPhone straight from a memory-mapped binary substrate: no parsing step, no JSON on the frame path, and a downtown viewport that touches one cell out of 1,375. The measurements are below. So are the gates that stand between this city and the planet, because the frontier is where the useful engineering lives.
Iterating on Navier-Stokes
Research claims inflate in a predictable way: a correct observation is written down, an implication that nobody typed as a claim gets attached to it, and five months later the two travel together into a deck. A claim ledger exists to make that impossible. This is a worked case from our own Navier-Stokes record, with the exact implication, the arithmetic that falsifies it, what the ledger now says, and the eight claim shapes any governance system for research has to hold without flattening them.
Iterating on Navier-Stokes
Since April 2026 we have used the Navier-Stokes regularity problem as a test article for 8DB, under one rule: every attempt is composed from the database's own primitives, and every place the database cannot express what the mathematics needs is written down as friction and becomes an engineering task. Six months of that produced an exact list of where a store built for regulated data falls short of research, and the fixes that shipped because of it: asymmetric uncertainty, typed proof obligations, write-time audited implications, derived trust tiers.
Post-Quantum Databases
The scanner says your database is fine. Your 2019 backup disagrees. Every snapshot, replica and archive is a copy of the cryptographic envelope as it stood the day it was taken, and it will still be readable the day that envelope's algorithm falls. Migration programmes that inventory the live system and stop have found the front door and left the warehouse open. Here is what a data system has to do so the copies migrate with the data.
Post-Quantum Databases
Derived data is data. The index, the embedding, the cache, the replica and the warehouse row are the same record wearing different structures, each with its own envelope and its own owner. Here is how to count the copies, why the other ten are the exposed ones, and what changes when the copies stop existing.
Post-Quantum Databases
Swapping the algorithm is the easy part. A post-quantum migration is an operating commitment across seven surfaces and three tenses, twenty-one cells in all, and most of the products on a 2026 shortlist fill one of them. 8DB fills the grid from one place. Here is the grid, what lives in the seams between the cells, and why the seams are the part to be afraid of.
Post-Quantum Databases
Every product that stores sensitive data for years will be asked the same question by federal, defence and regulated buyers: when an algorithm retires, how do the records already written move, and how do we prove it? 8DB is an embedded engine that makes the answer a property of your product instead of a project for each customer. Your interface, your customer relationship, our substrate doing the cryptographic, governance and provenance work you would otherwise have to build.
Post-Quantum Databases
Deployed systems, vehicles, sensors, field kits and disconnected sites hold sensitive data for years and cannot call a cloud key service or carry a hardware security module. The post-quantum transition still applies to them. 8DB establishes Category 5 keys in software, in process, on the hardware you already have, leaves no key database on the device, and pays the post-quantum signature once per batch instead of once per row.
Evaluation and Assurance
Merging records about people, companies and vessels from hundreds of sources is where intelligence analytics earns or loses trust, and most systems hide the merge behind a score. 8DB does the opposite: every merge decision carries its evidence, every conflict is preserved rather than resolved by fiat, every gap is stated as a gap, relationships stay whole, and a rerun over 4.28 million public sanctions records reproduces the same decisions byte for byte.
Post-Quantum Databases
Financial institutions hold decades of records that an adversary can copy today and read later. The G7 Cyber Expert Group roadmap names six activities for the transition; the two in the middle, execution and testing, are where stored data becomes the hard problem and where tooling built for network migration runs out. 8DB performs both as database operations and hands the examiner evidence that checks without trusting the operator.
Evaluation and Assurance
An agent reasoning at the tactical edge is only as trustworthy as the data it can reach. Put access control in a filter in front of plaintext and every mistake in the filter is a disclosure. Make clearance and policy inputs to key derivation and an unauthorised read yields ciphertext and nothing else. That is what 8DB does for sensemaking systems, and an independent evaluator can run the claim on an isolated host.
Fielded Systems
Edge nodes hold telemetry, geospatial data, imagery, relationships and time series, usually in five engines with five governance models and no connection to the services that were supposed to protect them. 8DB holds all of it in one sealed store, survives power loss, works with no reach-back, and moves sealed intelligence between nodes when a link appears. Each of those is a tested claim an independent evaluator can run on an isolated host.
Post-Quantum Databases
With the record cipher held constant, post-quantum key establishment adds zero per-record bytes. In the latest four-CPU evaluation, a 100,000-record migration met its one-second deadline for eligible reads and recovered from a halfway process kill. The evidence retains the passing run and earlier deadline failures.
Post-Quantum Databases
Inventory tooling tells an agency what is exposed. It does not move a decade of stored records under post-quantum keys, prove the move finished, or do it again when the next algorithm retires. 8DB is designed to run that transition as a database operation: an algorithm version on every record, two versions live at once, completion accounted per record, recovery under current authority, and a post-quantum envelope whose per-record cost is measured at parity with the classical one.
Fielded Systems
Predictive maintenance for a fleet lives or dies on whether a maintainer believes the alert, and that belief rests on the data layer, the part nobody designed. 8DB gives a sustainment program a substrate in which telemetry, subsystem relationships and events are captured on the platform, sealed at write, kept from leaving where they may not go, attributable to their source, and provable to a reviewer months later, with every one of those guarantees tested under independent evaluation.
Post-Quantum Databases
Executive Order 14412 asks for a cryptographic bill of materials, and every tool that produces one today scans from the outside. 8DB holds the answer on the inside: an algorithm version on every ciphertext, keys derived from each record's governance state, one post-quantum signature over every batch. The self-report turns that state into a CycloneDX 1.6 artifact a compliance team can file, per dataset, on demand, generated where the data lives. Early access is open to the teams who have to file it, to shape the fields that come next.
Evaluation and Assurance
Most product security claims arrive as documents. An evaluator should receive one pinned artifact, run it offline without source access, and get PASS, FAIL or INCONCLUSIVE against frozen baselines for a small set of bounded claims. That is how 8DB is packaged, and it is the shape every data product for high-consequence programmes should be held to.
Model-Independent Data
Before a model produces its first token, another system has already decided what reality it is allowed to see, and that system is fragmenting the evidence. 8DB lets one protected fact participate in relational, graph, vector, temporal, spatial and media computation without becoming a collection of disconnected truths. The result is AI context that arrives as queryable evidence: source, confidence, policy and history intact.
Post-Quantum Databases
Hold bulk encryption constant and the real economics of post-quantum storage appear. Moving key establishment to ML-KEM-1024 added zero bytes per record and no measurable per-record latency. Integrity is where the money goes: whether a post-quantum signature is repeated on every record or applied once to a native dataset boundary is the difference between 13.1 MB and 579.4 MB for the same 100,000 records.
One Protected Substrate
Graph, vector, document, time-series, spatial: the market sells a database per noun and leaves you to keep the copies aligned with ETL. 8DB treats each of those as a view over shared typed information, so the same durable record answers a graph question, a similarity question and a spatial question without being copied anywhere. That changes what a database should be compared against.
One Protected Substrate
Unified silicon gave platform designers control over how computation moves. A protected information plane gives them control over how meaning, evidence and authorization move: faster personal context, search that understands relationships, an assistant that remembers without exporting a life, and privacy controls that close the traces as well as the record. It arrives as software on hardware you already ship.