Now/ Updated 2 weeks ago
Research projects
James Brady is currently working on a number of different research projects. Which projects those are is not published here yet, and what ships from them will appear on this site with its numbers and its method when it is ready. This page is hand-edited rather than generated from repository activity, the age indicator beside it is computed from the date the file itself carries, and the open questions the whole site is still carrying are listed below.
James is currently working on a number of different research projects.
Which ones they are is not published here yet. What ships from them will be published on this site, with its numbers and the method behind them, when it is ready. Nothing goes up before that: a project described before it exists is a promise, and this site publishes results.
The rest of this page is the work log.
Open items
24 questions the site is still carrying, across 14 pages. Each one is a fact only James has. Where one of these is unanswered, the page it belongs to says the fact is not published rather than filling the hole.
of1.ai sign-in is currently broken, and the Cartographer feature is not working in production. Do you want that stated on this page, kept off it, or fixed before the page ships? Recommendation: fix or scope the page to the method and the open-source compiler, which are both provably live.
is there a second published cost-per-merged-PR figure you want on this page, from your own fleet rather than from the repo? If yes, supply the number, the window, and the repo it covers. Do not publish a fleet figure without your clearance.
Case study, The visibility platform
supply one outcome number with its method, for one anonymized client. Example shape: "an HVAC client, over 90 days, moved from N of 76 measures passing to M." Nothing goes on this page until you give the figure and the window.
Case study, The visibility platform
do you want a redacted screen recording of a real run for this page? The site brief asks for a 60 to 90 second watch-it-run clip on flagship studies. Approve the recording and the redaction list.
Case study, The visibility platform
clearance question. Any client willing to be named yet? If yes, that becomes a second, named case study and this one stays anonymous.
Case study, EEG meditation toolkit
label this "archived" or "dormant, open to revival"? Also: was this run on your own recordings from a specific headset, or on synthetic data only? If real recordings exist, name the device.
supply a performance number with its method. A Lighthouse or CrUX score for seopr1.com, with the date and the device profile. The site brief bans typed numbers, so this needs a real measurement you approve.
Case study, AI-readiness assessment
the README calls the 28 questions "research-based". Name the sources, or the page will say "questions drawn from an internal rubric" instead. Do not publish "research-based" without citations.
Case study, AI-readiness assessment
has this been run by any real organization? If yes, how many, and can any of them be described anonymously by industry? If no, the page says so plainly.
give the honest verdict line for this page. Which of these is getting real traffic, and what have you decided to kill? A thesis page with no result is just a list of domains. One sentence about what you learned makes it a case study.
confirm the exact set of "real applications". The inventory says 2 to 4. Name them.
Theory, Why should a useful-looking finding get more scrutiny, not less?
this method is currently an internal skill file. Do you want it published as a public repository, the way problem-solving-system is? If yes, it needs a client-safe pass first, because the working copy names two client engagements.
Theory, Can one person run a software team by splitting judgment from building?
the local hardened copy is unpublished. Publish a write-up of the v2.4 changes, link the upstream pull request, or keep both private? Recommendation: publish the write-up, link upstream, credit Dan by name in the first paragraph.
Theory, Do language models share a hidden vocabulary for relational states?
name one or two of the specific vocabulary words or states you kept seeing, if you are comfortable. One concrete example would make this page far stronger, and right now there is not one.
Theory, Do language models share a hidden vocabulary for relational states?
the originating ChatGPT conversations are on the MacBook Pro and did not sync. If you export them, this page can carry your actual first observations with dates. Optional. The page works without them.
Theory, Should proof of function come before proof of thoroughness?
this rule came from your own words on 2026-07-28: "ensure that we do not over engineer the testing. We want to focus on function. We can refine later. Great is the enemy of good." Do you want that quoted on the page as yours? It is the strongest line on it, but it is from an internal working session.
Work with me, Background screening for your hires, done clearly
state your exact relationship to Vuplicity in one publishable line, with the date it became true. Title, ownership, operator, adviser, whichever is accurate. Until you supply it, this page says only that a screening enquiry reaches you and that Vuplicity delivers the work.
Work with me, Background screening for your hires, done clearly
is there a screening figure you want published, with its method and window? Volume run, average turnaround, anything measured. Nothing goes on this page until you give the figure and the window it covers.
the human story. Two or three paragraphs, in your voice, covering: how you got here, what you were doing before this, and why you work this way instead of some other way. This is the single biggest gap on the page. Everything else below is process, and process without a person reads like a manual.
one sentence on what you are trying to build over the next few years. Not a mission statement. The actual goal.
a real photograph. The site brief lists a real photo asset as a build gate.
confirm this split is how you want it stated, and confirm the legal entity names and structure. Is Utlyze the parent, is New Reward a brand under it, or are they separate? This paragraph should match reality exactly, because it is the one a lawyer or a prospective client will read closely.
About, how the work gets checked
one measured throughput number would make this section land, with its method and window. Something in the shape of "N changes merged in a 7-day window, every one through the same required checks". You have measured this. Approve a figure and a window, or this section stays qualitative.
Contact, what makes a strong fit
what makes someone a strong fit to reach out, in one or two lines, in your words? The three bullets above are carried over from the current site. A real qualification in your voice would be stronger than a generic one.
- Source
- Every pending mark in the content source
- Method
- Scanned from the same bodies the pages render, counted at build