What changed.

Release notes page, and two gates that keep this page honest

A /release-notes page now carries the itemised ship log for the Theo surface — what changed, when, in one flat list — while this page stays the dated narrative of releases and corpus versions. Two mechanical gates ship with it: one holds the availability vocabulary (Theo is generally available; Majlis and Mizan are planned surfaces on the waitlist) so neither can drift into the other, and one reads the provenance date behind every published figure and this page’s own cadence, warning before either goes stale and failing when a provenance line is missing outright.

Read the release notes →

Substrate attribution corrected; portfolio canon resynced

How this site attributes the reasoning substrate beneath Theo was corrected, and the shared portfolio canon was resynchronised from source so all four MindHYVE grid sites describe the portfolio identically instead of drifting into four versions of one story. The engine named on any customer surface is Eve-Theology™ F5/reasoner — the layer beneath it is not disclosed here.

Privacy notice: what withdrawing consent actually does

Two corrections to what the privacy notice SAYS, not to what the platform does. Section 6 now describes the actual effect of withdrawing special-category consent, and states plainly that the record of the consent itself is retained. A notice that under-describes retention is not a smaller problem than one that over-describes it.

Read the privacy notice →

Annual prices, product naming, and localized canonicals

The advertised annual price for each plan now matches the amount checkout charges. The home-page product label described Theo in generic software terms and now names him as the Digital Employee he is — a scholar who returns an Evidence-Based Opinion with its chain of transmission, never an edict. Arabic, Urdu, Indonesian and Malay pages had declared the English page as their canonical and social URL, so a link shared in Arabic opened in English; each locale now points at itself. Eve, the shared Digital Employee who answers questions about MindHYVE products, is loaded from the shared origin rather than a site-local copy.

/capabilities — the whole census, on one page

The full customer-facing inventory is now published as a census taken from the code: 290+ capabilities across twelve groups, each group carrying its count and a named selection of what is inside it, plus the posture items counted separately so the headline stays a census of what a seeker can actually touch. Every figure is a floor rather than a rounded-up headline, internal operations tooling is excluded from all of them, and the census date is printed beside the numbers.

Read the capability census →

The capability census, published for machines

The same census is published machine-readably at /llms.txt — exact counts, the positioning language, the twelve-group map, and the terminology rules — so AI search and Eve quote the figures a reader sees instead of inferring different ones from prose. A usage gate that reads the BUILT HTML rather than the dictionaries also landed, closing the two i18n defects a file-to-file comparison is structurally blind to.

Read the machine-readable census →

One live source for every published figure

Member, citation and page counts had been published from three places at once: a live counter, a frozen copy baked into the hero and all five locale bundles, and a third copy in the press kit under a heading claiming it was live. The frozen copies had drifted 13% below the live value — visibly, on one screen. Every instance now reads one build-time source. Grading copy was corrected in the same pass: 480,334 of 617,912 cataloged narrations carry a grade (≈ 77.7%), and each surface now says catalog count where it means catalog and graded count where it means graded. The Majlis and Mizan pages stopped naming screens whose tables were dropped in May, dead mockups were deleted, and 34 dangling locale keys that were still shipping in the client bundle were retired.

Social cards, and a type gate on every pull request

Pages shared to social platforms and messaging apps now render a brand-forward preview card generated per page. The Astro/TypeScript diagnostic backlog was cleared to zero and made a blocking check, so a new type error fails a pull request instead of leaking into a green build.

Lead surfaces, disclosure accuracy, and a real merge gate

On mobile the cookie consent banner sat on top of the contact form’s submit button, so a visitor on a phone could complete the form and not be able to send it. Contact-form input is now masked from session replay, first-touch attribution was added on both lead surfaces, the AI-processor disclosure in the privacy notice was corrected to name the platform that actually processes requests, and the published billing contact was changed to one that reaches a mailbox a person reads. A pull-request validation pipeline and a brand-canon gate also landed: before them this repository had no required check at all, and two customer-affecting changes had merged with nothing watching.

Enquiries from campaign links stopped being dropped

Contact submissions arriving through a campaign link were rejected by the API and lost, because the form sent campaign parameters under names the API did not accept. They now submit and arrive.

Build identity on every deploy

The site publishes its own build identity, so any reader — or any engineer chasing a report — can establish which build a page was served from instead of guessing whether a deploy landed. Duplicate builds triggered by the same commit are also collapsed, so a deploy queue reflects real changes.

Eve replaces the marketing launcher

The site’s own conversation launcher was replaced by Eve — MindHYVE’s Digital Employee, specialised on this site for theology questions about the platform. The division of labour is deliberate: Eve answers about TheoAI, Theo answers from the corpus, and Theo stays exactly where he was at chat.theogrid.ai.

Home page leads with the evidential substrate

The home hero was rebuilt to lead with what actually grounds an answer — the corpus, the parsed chain, the narrator graph — rather than with a claim about the platform. The narrator graph itself was kept: it is the clearest single picture of what an isnād is.

Bias architecture on the record, single-typeface system, site audit

The Eve-Genesis technology page now carries a “how we handle bias” section in the portfolio voice: you cannot train bias out of a model — you separate the reasoning from the knowledge from the jurisdiction, so the bias becomes readable, auditable, and governable (population-free reasoning plus runtime steering, never the claim of an “unbiased” model). Typography unified to a single typeface — IBM Plex Sans for display, body, and italic, with Amiri for Arabic. A full page-and-folder audit of theogrid.ai wired every standalone section into the menu and footer so nothing is orphaned. Headcount updated to 170 and Theo’s served-question figure to 467,000+; trademark and licensing contact corrected to MindHYVE.ai™ / licensing@mindhyve.ai.

Cookie consent banner + Privacy controls

Added a GDPR/ePrivacy-compliant consent banner across www.theogrid.ai and ipsc.theogrid.ai. Microsoft Clarity, Application Insights, Meta Pixel, and our internal event tracker are all gated on explicit consent. Anonymous browsing is honoured.

Investor-facing About page rebuild

About page deployments section reframed as the parent-company (MindHYVE.ai) track record with corrected product attributions: ArthurAI ULE/CDS at CNU, ArthurAI VLE/CLE at LaGrange, ChironAI global launch, and the OUK/CIAI/Africa HQ partnerships. Press list now includes verifiable links to each PRNewswire / GovTech / Bastille Post source.

TheoAI launches publicly

chat.theogrid.ai opens to the public. 15 free questions per day for anonymous visitors; full corpus access on subscription tiers (Plus, Tahqīq, Institute).