Everything that is
actually running.

The problem. AI regulation arrives from five directions at once — FTC enforcement, state and federal legislation, bar ethics guidance, ABA guidance, and the EU AI Act — and no single feed covers them together. Anyone tracking it by hand is reading five newsletters and hoping none of them lapsed.

What it does. Runs a scheduled sweep every Monday, files what it finds into those five categories, and publishes the result as a dashboard and a JSON endpoint. The API returns a collection-health block alongside the documents: last sweep time, age in days, failed-search count, newest item date, and a stale flag that trips after two missed cycles. A stale register is shown as stale rather than presented as the current picture.

Built on. Vercel serverless functions with an Upstash Redis store for the collected corpus and sweep telemetry. No account, no database of readers.

What it does not establish. This is a collector, not a verifier. Results come from automated search, and the corpus includes secondary reporting about enforcement actions as well as primary documents. Nothing in it is a legal conclusion or a substitute for reading the source.

The problem. Brand and campaign teams treat a film or television moment as a demand signal long before anyone has asked what would make that inference testable, or what governance questions an AI-assisted read of culture raises.

What it does. Tracks film, television and cultural moments alongside consumer style and spending signals, and attaches an AI governance risk score to each read so the inference and its reliability travel together rather than the score arriving alone.

Built on. A Next.js pages-router application deployed on Vercel.

What it does not establish. Exploratory. The signals are hypotheses about consumer behaviour, not measured outcomes, and the risk scoring is a framing device rather than a validated instrument.

The problem. Controlled-substance scheduling moves through the Federal Register in several document types at once, and the ones that decide anything — final rules, proposed rules, temporary orders, quotas — arrive mixed in with registrations and routine notices. Reading it raw means sorting signal from filing traffic by hand.

What it does. Pulls DEA documents from the Federal Register API over a rolling 180-day window and sorts them into high-signal actions, final rules, proposed actions open for comment, and quota and temporary orders. Registrations and general notices stay available as reference categories, and the counts across views reconcile to the same document set rather than being computed separately.

Built on. Scheduled collection against federalregister.gov, the primary source, with a source-health panel that reports the refresh cadence so a stale feed shows as stale.

What it does not establish. Automated document discovery, not legal analysis. It tells you a document exists and what type it is; it does not tell you what it means for any particular substance, registrant, or programme.

The problem. When a court sanctions a lawyer for filing fabricated citations, the coverage travels much further than the order. Practitioners need the actual authority — the ABA opinion, the state bar guidance, the standing order, the published sanctions opinion — separated from commentary about it.

What it does. Holds a hand-verified register of primary authorities on lawyers' use of generative AI: ABA Formal Opinion 512, four state and city bar opinions, two Middle District of Florida standing orders, and a published Ninth Circuit sanctions opinion. Each carries its jurisdiction, authority type, verification date and a link to the official source. A separate weekly sweep searches official domains for new candidates; anything it finds is filed as an unverified candidate and stays that way until a reviewer opens the document and confirms the characterisation. The two tiers are never merged.

Built on. Vercel serverless functions over Upstash Redis, with the verified register held in source and the discovery layer kept strictly downstream of it.

What it does not establish. The eight verified authorities were confirmed on 17 August 2026. The discovery sweep has not completed since 10 August 2026 and the site reports itself stale; the verified register is unaffected by that, but recent guidance may be missing. Not legal advice.

The problem. Arguments about platform moderation of drug-related speech run on a mix of evidence types that get quoted interchangeably: an Oversight Board decision, a regulator's inquiry, a platform's own statement, and an advocacy group's allegation are not the same kind of fact, but they are cited as though they were.

What it does. Keeps a source-linked register in which every record states four things separately — the claim, what its source actually establishes, the evidentiary boundary, and the verification date. An overturned Oversight Board decision is recorded as a documented reversal in an individual case, with an explicit note that it does not establish a platform-wide pattern, a motive, or a legal violation. Review state is kept in the visitor's browser.

Built on. A static front end over a versioned JSON dashboard endpoint, with a build-time validation script and a test suite. No account, no analytics, no submission endpoint; browser-local state only.

What it does not establish. Bounded by design — it covers the documented public record, not the full universe of moderation decisions. Priority flags are local triage, not findings.

The problem. A founder does not have a legal department, and the legal exposures that matter to an early company are scattered across agencies that each publish separately. The usual answer is a long checklist that nobody owns, so nothing on it ever becomes an action.

What it does. Keeps a selected register of United States startup legal exposures, each one linked to the official source and labelled by source type. Every record can be moved through a review status — unreviewed, watching, ask counsel, closed — so the register becomes a set of owned decisions rather than reading. Records can be filtered by operating profile and product exposure, saved, and exported as JSON.

Built on. A static workspace with all review state held in the browser. No account, no tracking, and nothing about which records you read leaves the machine.

What it does not establish. Deliberately selected rather than comprehensive, and educational issue spotting only. It is not legal advice, creates no attorney-client relationship, and the register does not attempt to cover every exposure a company may have.

The problem. Psychedelic regulation moves through state programme portals and rulemaking pages that publish on no fixed schedule. Most trackers in the space blur an official rule change together with a press release or an advocacy claim.

What it does. Watches six verified official source endpoints — Oregon Health Authority psilocybin services, Colorado natural medicine, and four others — on a weekly cycle. Every candidate a query returns is put through a source policy before it is shown; on the most recent sweep, seven candidates passed and forty-one were rejected as noise. The page reports per-category sweep state, how many sources were healthy, degraded or failed, and how many days old the last successful sweep is.

Built on. Vercel serverless functions with an Upstash Redis store and an explicit source-policy module that every candidate must clear before display.

What it does not establish. Screened candidates are unreviewed. A record in the source registry means the endpoint is monitored, not that something new was published there. Not legal advice.

The problem. Drug-policy transitions are argued from whichever dimension the arguer cares about — health, or crime, or revenue — and the dimensions that were not measured quietly become the ones that did not matter.

What it does. Builds a source-anchored evidence record across six dimensions for a given drug-policy or regulatory transition, so a claim about one dimension can be read against what is and is not documented in the others.

Built on. A Next.js application on Vercel.

What it does not establish. It organises published evidence. It does not generate findings, and an absence of evidence in a dimension is recorded as an absence rather than as a null result.

The problem. Claims about screen media reshaping how people see themselves circulate as either dismissal or alarm, with the actual evidence rarely laid out in a form where a reader can see how far it reaches.

What it does. Sets out a sourced hypothesis about how screen worlds may become self-relevant, presented as an interactive research note with an evidence ledger attached and its limits stated in the body rather than in a footnote.

Built on. A Next.js pages-router application on Vercel.

What it does not establish. A hypothesis with an evidence ledger. It is not a finding, and it reports no original study.

The problem. Six audiences — clinicians, employers, families, journalists, policymakers, patients — need the same pharmacology and the same law, but each needs it framed for a different decision, and a single undifferentiated explainer serves none of them.

What it does. Covers pharmacology, law and policy for psychoactive plant medicine across six audience paths, over a 222-source reference base. A verification record is committed alongside the site, so the citation set can be re-checked by someone who does not trust the author.

Built on. A static site with a committed verification manifest, served from Vercel.

The problem. The popular history of psychedelics is built on a handful of dramatic claims that have quietly failed replication, and they are repeated because the confirmed record and the contested one are told in the same voice. The 60,000-year Shanidar Cave story is the clearest case: the pollen it rests on is now attributed to burrowing rodents, and the assemblage contains no entheogen at all.

What it does. Assembles archaeology, neuroscience, legal history, phenomenology and religious use into one atlas anchored on the oldest chemically confirmed use of a psychedelic — peyote from Shumla Cave, radiocarbon dated to 3780–3660 cal BC and shown by GC-MS to contain mescaline. Older claims are carried where they are serious, but held on an explicit evidence ladder rather than folded into a single headline figure. A separate evidence register and governance page record how each rung is assigned.

Built on. A static single-page atlas with a committed evidence register and a small governance endpoint. No build step; served from Vercel.

The problem. Thirty-four years of one researcher's recorded public output — talks, interviews, papers, broadcasts — is scattered across platforms that come and go, and there is no way to see the shape of the whole or to know which references still resolve.

What it does. Catalogues 132 entries spanning 1968 to 2026, each carrying a verified link status so a dead reference is visible as dead rather than silently broken. Offers local search across the catalogue, an annotation aid whose workings are shown rather than asserted, and a full archive explorer.

Built on. A single static HTML document served from Vercel. No build step, no dependencies, no account.

What it does not establish. A catalogue of recorded output. It does not reproduce the works, and it is not an authorised archive.

The problem. Oregon and Colorado wrote genuinely novel statutes and rules, and the people most affected by them — facilitators, applicants, small operators — are reading them without counsel and without a plain-English intermediate layer that still points at the actual authority.

What it does. Publishes 143 plain-English articles on Oregon and Colorado law, facilitator requirements, and psychedelic business questions, each carrying its sources, with a separate sources index across the whole library and a per-article manifest.

Built on. An Astro-built static site with prebuilt output committed to the repository, served from Vercel.

What it does not establish. Plain-English explanation of published law. Not legal advice, and not a substitute for the statute or rule it describes.

The problem. Directories in an emerging regulated field are read as endorsements. Whoever assembles one is assumed to have vetted licensure, quality and financial stability, and almost never has.

What it does. Indexes 175 organisations with a category, a subcategory, a link status and a record-review date. Pricing, promotional claims, credentials, rankings and detailed service descriptions are deliberately withheld until claim-level provenance exists to support them. The API states the count, the review dates and the withheld fields; the current sweep flags 14 listings whose links no longer resolve.

Built on. A static front end over a serverless listings endpoint with a static fallback, plus a contract test suite. Methodology and corrections pages ship with it.

What it does not establish. Inclusion is not vetting, endorsement, licensure verification, or any finding about quality, safety, financial stability or legal compliance.

The problem. A reading list in a contested field is a set of assertions about what is worth reading, usually made without showing where anything came from or whether the rights to use it were ever checked.

What it does. Publishes 454 editorial objects — quote cards, then-and-now comparisons, flashcards, mini-essays, educational modules, character profiles, passage spotlights, companion reading lists, a scientific annotation layer, source packs — each drawn from a 1,664-record provenance ledger with per-item source visibility and rights-review notes. A research desk lets a reader hold a saved query across a session.

Built on. A static site with a committed provenance directory and a governance page, served from Vercel.

What it does not establish. The published set is a subset of the ledger by design; items without settled rights review stay unpublished. It indexes and annotates literature, it does not reproduce it.

The problem. Custom AI projects get quoted as one number, even though discovery, research, architecture, implementation, testing, documentation, launch and maintenance are different kinds and amounts of work. That opacity hurts the buyer and the builder equally.

What it does. Walks six stages — project brief, editable labour architecture across eight workstreams, model and API workloads, infrastructure, one-time and recurring costs, named uncertainty drivers — and returns cost composition, sensitivity drivers, build and monthly ranges, and a first-year total. Scenarios save locally, share by URL, print to PDF, and export as CSV and JSON. The model-price catalogue is dated and editable rather than hidden behind a project-type selector.

Built on. A static browser application with local persistence and URL-encoded scenario sharing, served from Vercel.

What it does not establish. Outputs are the user's own planning scenarios. They are not quotes, and they are not independently validated market benchmarks.

The problem. Attention signals from screen culture get converted into consumer-behaviour conclusions with no step in between where someone writes down what would have to be true for the inference to hold.

What it does. Turns a screen-culture attention signal into an explicitly testable consumer-behaviour hypothesis, and states a use boundary on the same page.

Built on. A Next.js pages-router application on Vercel.

What it does not establish. This is a single-screen prototype of roughly 130 words. It is listed as one, not as a product. The hypotheses are untested.

The problem. A creator using AI runs into disclosure duties, likeness rights, music licensing, platform policy and jurisdiction all at once, usually without knowing which of the five is the one that will actually bite.

What it does. Asks seven questions and returns the issue areas they implicate across creator rights, AI use, disclosure, likeness, music, jurisdiction and platform review.

Built on. A Next.js pages-router application on Vercel.

What it does not establish. Educational issue-spotting. Not legal advice and not a jurisdiction-specific answer.

The problem. Governance questions on an AI-enabled campaign surface after it ships, when the answer is an apology rather than a change.

What it does. Runs a browser-local checklist over a cultural or AI-enabled campaign, linking each question to its official source and stating where the checklist stops.

Built on. A Next.js pages-router application on Vercel, with no submission endpoint.

What it does not establish. Issue-spotting. Not a legal review and not a clearance.

The problem. Creative teams adopting synthetic media move faster than the decisions about provenance, disclosure and what happens when something goes wrong — and those decisions are then made under incident conditions.

What it does. A browser-local workbench for mapping synthetic-media, creator-rights, disclosure, provenance and incident-response decisions before they are needed, so the incident plan exists in advance of the incident.

Built on. A Next.js pages-router application on Vercel.

What it does not establish. A governance mapping tool. Not a legal review.

The problem. Pre- and post-questionnaire designs need to link a person's two responses without holding their identity, and the linking mechanism is the part that usually gets hand-waved in a demonstration.

What it does. Demonstrates pseudonymously linked pre and post questionnaire records held entirely in the browser, so the linkage design can be inspected rather than described.

Built on. A static browser application with local persistence, served from Vercel.

What it does not establish. Explicitly unvalidated. It does not measure effectiveness, knowledge gain, behaviour change or clinical outcomes, and no instrument in it has been normed.

The problem. Organisations deploying AI agents are asked for an inventory, a risk assessment and an evidence trail at the moment a regulator or a customer asks, which is the worst possible moment to start assembling one.

What it does. Works through AI system inventory, risk triage, controls, evaluation, incidents and human oversight, and exports the result as an evidence pack. Everything stays in the browser.

Built on. A static browser application with local persistence, plus privacy and terms pages, served from Vercel.

What it does not establish. Local-first issue-spotting. Not a compliance certification and not a legal assessment.

The problem. Families and schools are asked whether children should use AI as though the tool, the task, the child's age, the data practices and the level of adult oversight all raised the same question. They do not, and the general answer is useless for the specific decision in front of someone.

What it does. Evaluates one specific child-facing AI use across five age-aware stages: context, the role AI will play, what cognitive work it may substitute for, adult involvement, consequence level, and data and human safeguards. The decision logic is inspectable rather than hidden, and the output names the developmental capacities to protect, the questions to put to a vendor or school, and the signals to monitor.

Built on. A static browser application with local persistence and dated source links, served from Vercel. The decision model draws on the AAP 2026 state-of-the-art review, UNICEF child-centred AI guidance, UNESCO guidance for education, US Department of Education FERPA guidance, the FTC's amended COPPA Rule, and the NIST Generative AI Profile.

What it does not establish. It does not diagnose children, certify products, determine legal compliance, or claim that long-term neurological effects are settled.

The problem. Neuro-aware design claims are easy to make and rarely tested. A building gets described as calming, and nothing in the process ever asks what observation would have shown otherwise.

What it does. Structures spatial observation into evidence-aware design hypotheses, runs an accessibility review, prioritises the resulting changes, and sets up post-occupancy measurement so a hypothesis can be checked after the space is occupied.

Built on. A static browser application with local persistence, plus privacy and terms pages, served from Vercel.

What it does not establish. A structured observation and hypothesis tool. Not an accessibility certification and not a code compliance review.

The public link is held pending the release review recorded against this build. The deployment answers, and its content is described above, but it is not promoted from this site until that review clears.

The problem. Advice about environment and routine arrives as universal prescription, when the thing that varies most is the person — and there is rarely any mechanism for someone to find out whether a change actually did anything for them.

What it does. Runs a five-minute sensory profile, then structures a loop the site states plainly as association, then hypothesis, then test: record the environment, the starting state, how it felt, and optional product context, and let the patterns accumulate across entries rather than asserting an effect up front.

Built on. A single-file, dependency-free browser application served as a static site from Vercel, with a research dossier held alongside it.

What it does not establish. A research preview with a waitlist, not a released platform. It is not a medical device, not a diagnostic, and it makes no clinical claim.

The problem. The NSAG assessment modules each collected enquiries on their own, so there was no single place to see which modules were drawing interest, from what kinds of organisation, or whether a tier was producing anything at all.

What it does. Aggregates enquiries across the NSAG modules into one dashboard, filterable by module and by organisation tier, with totals for leads, active modules and organisations represented, and a CSV export.

Built on. An admin-key gate in front of the dashboard; the lead data is not served until the key is accepted.

What it does not establish. This is internal tooling, not a public product. It is listed here because it is part of the estate and reachable, not because there is anything on it for a visitor to use — without the key it shows only the sign-in.

The problem. Independent professionals accumulate disconnected notes, task managers, client records, AI conversations and automations. The tools are not the constraint; deciding what belongs where, what happens next, and how the system stays current is, and that work never leaves the operator.

What it does. A six-stage practice-architecture workbench: a maturity and friction map, editable workflow and tool inventories, governed AI role cards, a weekly review builder, a prioritised 30-day implementation sequence, and five portable data schemas. Includes an example mode, browser persistence, and Markdown, JSON, CSV and print exports.

Built on. A static browser application with local persistence, served from Vercel.

What it does not establish. A planning tool, not a certification and not an autonomous operating system.

The problem. Documents now arrive containing authorities that do not exist, sources that do not support the proposition attributed to them, and confidence that outruns the evidence. A spot check cannot show which claims were actually reviewed, which is the thing a reader needs to know.

What it does. Turns claims and citations into a review ledger across five stages: review brief, claim ledger, source register, claim-level findings, and disposition. It tracks claim type, materiality, retrievability, source type, support, contradiction, currency, attribution, conflicts and reviewer responsibility separately, rather than inferring validity from the presence of a citation marker. Outputs a release packet as Markdown, CSV, JSON or print.

Built on. A static browser application with local persistence, served from Vercel.

What it does not establish. The score measures documentation completeness, not truth and not source authentication. Legal, clinical, scientific and regulatory conclusions still need the appropriate qualified reviewer.

The problem. Athlete welfare policies are written at the level of principle and then have to survive contact with scheduling, recovery, safeguarding and a young person's actual ability to raise a concern. The gap between the policy and the week is where the harm lives.

What it does. Runs a programme and decision scan across scheduling, recovery, safeguarding, environment, athlete voice and escalation, then produces an owned action register and a 90-day governance roadmap with named responsibility and an independent response and review path.

Built on. A static browser application with local persistence, plus privacy and terms pages, served from Vercel.

What it does not establish. Issue-spotting for a governance conversation. Not a safeguarding audit and not a clinical or legal determination.

The problem. Neuro-health questions that cross life stages, cannabis, psychedelics, medications and symptoms tend to get one appointment's worth of attention, and the patient arrives without the specific questions that would make that appointment useful.

What it does. Helps a person assemble what they are actually asking, what is known about it, and what to raise with a clinician, and produces a prepared set of appointment questions. Everything stays in the browser.

Built on. A static browser application with local persistence, plus privacy and terms pages, served from Vercel.

What it does not establish. Evidence orientation and appointment preparation. Not medical advice, not a diagnosis, and not a treatment recommendation.

The public link is held pending the release review recorded against this build. Health-facing guidance is promoted only after that review clears, whatever the deployment currently returns.

The problem. Lawyers, physicians, executives and educators are each being sold generic AI literacy, when what changes between them is the professional duty that the technology runs into.

What it does. Sets out a prelaunch AI-literacy programme for those four professional audiences and the shape of what each one would cover.

Built on. A Node-tested static site on Vercel.

What it does not establish. Prelaunch. Checkout is closed and personal-data inquiry collection is disabled, so nothing on the site can be purchased or signed up for yet.

The problem. Most people building with AI alone stop, and they stop for a reason that has nothing to do with the technology: nobody is expecting them at a particular hour.

What it does. Runs a daily body-doubling room for people building with AI — show up, build alongside others, leave with something shipped.

Built on. A static landing page with a serverless lead endpoint writing to Upstash and Slack, and an evidence file in the repository mapping every claim on the page to a primary source.

The problem. Small business owners are told AI will help and handed a blank prompt box. For someone with payroll to run on Friday, an open-ended assistant is not an opportunity, it is one more thing to work out. The gap is not enthusiasm; it is a first workflow concrete enough to finish in an afternoon and useful enough to repeat the following week.

What it does. Says what the tool actually is before recommending anything, names the systems it connects to, then walks three starter workflows an owner can pick from on the day: a Monday cash-and-invoices reset, a month close, and a morning brief. It ends where most quick-starts begin — what to connect first, and a short, unglamorous section on safety and cost.

Built on. A static single-page site with an installable web manifest, served from Vercel.

What it does not establish. Independent. Not affiliated with, endorsed by, or produced for Anthropic.

The problem. Institutions that need a degree programme in a fast-moving field face a build-or-wait choice, and building one from nothing is a multi-year commitment they usually cannot make in time.

What it does. Offers two delivery-ready programmes — a B.S. in AI and Machine Learning and a B.S. in Cybersecurity and Critical Infrastructure Protection — for institutional licensing, alongside a commissioned-development path.

Built on. Static pages generated by a small Node build script from a shared head, navigation and footer, on the shared estate design system, served from Vercel.

The problem. Competitive robotics has real on-ramps in Hawaiʻi, and no plain guide to them. The practical consequence is that the families who find their way in are disproportionately the ones who already knew someone inside, which turns a programme meant to widen access into one that quietly narrows it. The missing artefact is small: a page answering the questions a parent actually has, in the order they have them.

What it does. Explains what competitive robotics actually means before asking a family to commit to it, then covers how to start, roughly what it costs, where to look on Oʻahu, and — the section most guides skip — what families commonly get wrong on the way in. It closes with local examples rather than mainland ones.

Built on. A static single-page site with an installable web manifest, served from Vercel.

The problem. Most AI training teaches people to use a tool. Almost none teaches them to interrogate its output before relying on it, which is the skill that fails first in practice.

What it does. Runs free applied non-credit courses on responsible AI work — responsible AI verification, claims and sources and citations, data before prompts, and human oversight that works — alongside a field brief library, applied field tools, and progress that resumes where a learner left off.

Built on. A static multi-page site with client-side course and tool scripts, served from Vercel.

What it does not establish. Non-credit. No certification, no accreditation, no continuing-education credit.

The problem. Young inventors are told to protect the idea and left to work out what that means from sources written for people who already have counsel. The costly mistakes happen early and quietly — usually a disclosure made before anyone explained what disclosure does — and by the time someone can afford advice, the option they needed has often already closed.

What it does. Explains what a patent actually is, sets out four gates an idea has to clear to be patentable, walks the first steps, and shows how to run a free search without paying anyone. It spends a section killing the myths that cost inventors the most — chiefly about disclosure — and points at free Hawaiʻi help, including a number to call.

Built on. A static single-page site with an installable web manifest, served from Vercel.

What it does not establish. General information prepared with a law firm. Not legal advice and not an attorney-client relationship.

The problem. People are asked to work with regulation without ever being taught the underlying skill: telling a binding authority from persuasive commentary, checking whether it still applies where and when they are, and knowing when to escalate.

What it does. Teaches classifying authority, tracking jurisdiction and currentness, comparing regimes, and documenting escalation, as an interactive lab rather than a reading list.

Built on. A static site with its own academy stylesheet and script, plus privacy and terms pages, served from Vercel.

What it does not establish. Regulatory literacy training. Not legal advice and not jurisdiction-specific guidance.

The problem. Law students are told to network and given no system for it, which produces either nothing or a volume of undifferentiated outreach that works against them.

What it does. Presents a LinkedIn networking system for law students as a structured product, with a repository-level commerce registry kept in the open alongside the site.

Built on. A static site served from Vercel, with the commerce registry committed as a Markdown record.

The problem. AI literacy material is either marketing for a product or an abstraction with nothing a reader can go and check. The gap is a guide that names its sources and survives someone actually following them.

What it does. Runs a free, source-forward guide across twelve published routes — history, concepts, techniques, use cases, tools, playbooks, templates, the frontier, and a library — over 237 resources the site states are each verified and linked. It covers 76 years of history and 17 sourced turning points on its timeline, requires no account, and keeps optional reading progress only in the visitor's browser.

Built on. A static site on Vercel, with a machine-readable deployment registry recording that the superseded z2f-live hostname now only redirects here.

The problem. Supervised legal work is learned by doing it under someone who has done it, and the structures that make that transferable — reusable work formats, review points, what supervision actually checks — are rarely written down anywhere a student can see them.

What it does. Presents a public educational prototype of supervised legal workflows and reusable work structures.

Built on. A Node-tested static site on Vercel.

What it does not establish. Paused. Matching, applications, payments, analytics and inquiry APIs are all inactive, so nothing on the site can currently be applied to or purchased.

The problem. Part-time online law students assemble critical information across institutional pages, emails, calendars, informal knowledge and third-party sources — and a directory without freshness and correction controls makes that worse, not better, because it adds a source that looks official and is not.

What it does. Runs a mobile-first hub with a newsletter, an all-purpose guide, a filterable opportunity tracker and seven awareness campaigns. The part that matters most is the machinery around them: a stated source-verification policy, a correction pathway anyone can use, privacy rules, freshness controls, and automated link checking and secret scanning in continuous integration. A resource a student relies on for a deadline has to be maintainable by one person who also has coursework, so the maintenance burden was designed for before the content was.

Built on. A static multi-page site with automated link checking and secret scanning in CI, served from Vercel.

What it does not establish. Maintained by a student leader for the FlexJD community. Not an official statement, publication or record of Northeastern University or its School of Law; requirements, policies, deadlines and contacts must be verified with the original source.

The problem. A gallery rebuilt from an owner archive has to prove it is the same gallery — same revisions, same media, same approvals — or it is just a new site making claims about an old one.

What it does. Presents 14 complete story drafts about the networks, instruments, standards, archives and living systems Hawaiʻi gave the world, with 140 visual slides, 14 interactives and nine platform drafts per story, rebuilt from the verified owner archive with all 14 immutable public revisions, 18 controlled media files, frozen release bundles and approval records preserved.

Built on. A build-and-test pipeline over a committed public-record snapshot sufficient to rebuild every reader-facing page, with verified binary originals still segregated in the private owner archive.

What it does not establish. The collection is built for structured review and the site states that publication is not authorised. These are story drafts under review, not published articles.

The problem. A research programme that publishes for a public audience and runs a private review process at the same time has to keep the two apart in a way that survives a build, not just a policy.

What it does. Publishes four canonical editions — including The $2,000 Video, The Person Inside the Asset, and The Asset's Afterlife — with five interactive evidence experiences and a sanitised public Source Desk.

Built on. A privacy-separated reader application with build-time public-boundary and artifact-integrity checks, so the private studio, review records, unreleased audio and uncleared assets cannot reach the deployment. Served from Vercel.

The problem. Independent professionals are assessed through institutional proxies — employer, title, credential, network — while the work, capability and relationships that actually constitute their practice stay illegible.

What it does. Runs the public face of a research programme studying how individual capability becomes durable, portable and ownable: an Observatory, the Ownership Index instrument, Portfolio Professional research, a Question Lab, and partnership and applied-inquiry pathways.

Built on. A research platform with evidence-governed measurement instruments and longitudinal infrastructure, served from Vercel.

What it does not establish. The Ownership Index and the Portfolio Professional instrument are candidate instruments. Neither is validated, normed or activated, and the programme's own gates say so. Participant data and research operations stay private.

The problem. Event coverage produces a lot of claims and very little that can be checked afterwards, because the evidence stays in the reporter's notes and the notes are never published.

What it does. Publishes a searchable, city-filterable atlas of 48 source-backed investigations from the Netherlands 2026 event work. Each package opens as a public investigation page carrying its evidence, its limitations, its visual and its source links.

Built on. A production system that publishes the public atlas while keeping documentary media, rights records, internal source notes and human-QA evidence access-controlled. Public release does not convert an unresolved internal approval into a verified fact.

What it does not establish. The public atlas covers the Amsterdam and Eindhoven corpus only. Private rights and production records remain private.

The problem. Organisations are pushed from a vague intention to use AI straight into vendor selection. The step that gets skipped is deciding what the bounded problem is, what the non-AI alternatives are, who is affected, who owns it, and in advance, what would make a pilot stop.

What it does. Carries the public service catalogue, learning products including a twelve-issue Decision Desk and a citation verifier, machine-readable catalogue and feed routes, an offline recovery shell, and four browser-local decision aids: the Opportunity Studio at /studio, which walks twenty-three steps to map the work before any tool is chosen, and three at /tools — a decision record that preserves alternatives, owner, uncertainty, stop conditions and review triggers; a vendor comparison that tests the current process and two candidates against five non-negotiable gates plus your own weighted criteria; and a pilot designer covering hypothesis, baseline, measures, authority, affected-party input, stop and rollback.

Built on. A Next.js application on Vercel with a clinic inquiry form that fails closed unless its mail and anti-abuse environment variables are configured. Decision tools keep working data in the visitor's browser.

What it does not establish. The platform demonstrates the tools, not completed engagements. The Opportunity Studio describes itself on the page as a prototype whose answers stay in the browser. Nothing here is professional advice, a compliance certification, or an authorisation to deploy.

The problem. A company expanding across several jurisdictions accumulates legal and regulatory exposure faster than anyone can brief it, and the items that need a lawyer get mixed in with the items that do not.

What it does. Tracks legal and regulatory signals relevant to an international expansion across the US, UK, Germany and Poland, scoring each by business impact and likelihood, and flagging on the record which items require outside counsel confirmation before anyone acts on them.

Built on. A Next.js application on Vercel over owned Supabase infrastructure.

What it does not establish. That any of it was commissioned, adopted, or checked by a lawyer — 18 of the 27 signals are flagged for attorney confirmation precisely because the system does not supply it.

The problem. Brand and creator strategy in regulated markets — cannabis, wellness, health, legal, education, AI — runs on intuition and then meets disclosure rules that were written for exactly that situation.

What it does. Sets out an evidence-graded method for brand, creator and affiliate strategy in trust-sensitive markets, with recommendation confidence levels attached to each output and a governance and disclosure layer rather than a general claim of compliance.

Built on. A static multi-page site on Vercel, with the method and regulatory references dated on the page.

What it does not establish. Marketing method grounded in FTC endorsement guidance. Not legal advice and not a compliance opinion.

The problem. Identifying the moment a scientific or regulatory situation turns is a judgement call made repeatedly and rarely written down, which makes it impossible to audit and impossible to hand over.

What it does. Sets out an interactive public-source decision room for a proposed pilot: the timeline, research universe, account portfolio and governance workflow it would use.

Built on. A static interactive site on Vercel with downloadable brief, report, deck and research registers.

What it does not establish. Proposed scope, not pilot results. It was not commissioned, authorised, reviewed, approved, endorsed or adopted by the organisation it addresses.

The problem. Governance advice about human-facing systems is usually written as policy and handed over, with no account of how it becomes a decision anyone has to make on a Tuesday.

What it does. Presents evidence-led governance modules and human-supervised institutional advisory work, and is the current home of the work that the fifteen retired M-series prototypes were the first attempt at.

Built on. A static site on Vercel, with earlier full versions archived in the repository rather than deleted.

What it does not establish. The current site is the maintained surface. The fifteen numbered prototype modules listed further down this index are retired and are not part of it.

The problem. Most engagements start from a solution someone has already chosen, which means the actual question never gets examined.

What it does. A single-page front door that asks for the question rather than the brief, and sets out how the work proceeds from there.

Built on. A Next.js application on Vercel.

The problem. A portfolio and an evidence registry answer different questions — what is worth reading, versus what exists and still runs — and collapsing them into one page serves neither.

What it does. Runs as the evidence registry side of that pair: what has shipped, with boundaries visible.

Built on. A Next.js application on Vercel.

What it does not establish. Overlaps deliberately with this site. This site is the curated selection; that one is the registry.

The problem. A technical stack can be local, open and place-based and still leave the questions of authority, knowledge boundaries, accountability, withdrawal and refusal with nowhere to live.

What it does. Demonstrates a possible decision protocol, a living system record, a bounded pilot path and an optional learning translation, organised around the governance questions a community-authorised process would still have to answer.

Built on. A prototype application on Vercel, grounded in the recipient organisation's own published descriptions of its work.

What it does not establish. An independently prepared working proposal. It is not the recipient's product, policy, endorsement, approved protocol or community consent, and the substantive values and permissions would be theirs to define, not the author's.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered governance & accountability.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered clinical governance.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered education standards.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered public-sector assurance.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered vendor assurance.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered measurement & evaluation.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered implementation readiness.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered risk & compliance.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered privacy & data protection.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered ai governance.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered cybersecurity.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered quality systems.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered workforce competency.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered accessibility.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.

The problem. One of fifteen prototype modules that split institutional governance into separate assessable domains. This one covered research integrity.

What it does. Serves a retirement notice. The module's assessment is no longer available and the page says so; the deployment stays up so that links to it resolve to an honest statement rather than a 404 or, worse, a working-looking form.

Built on. A static page on Vercel.

What it does not establish. Retired. It is not maintained, it should not be treated as a current product, and the current advisory work lives on the NSAG site instead.