Apa yang Berubah di TheoAI

The portfolio band is on /about, and EliAI is the fifth Operating System

The component naming the sibling Agentic Operating Systems existed in this repository and was imported by no page, so its markup appeared in no build. TheoAI was therefore the one grid site that never rendered the shared portfolio statement — a component that exists and is rendered nowhere reads as done from the file tree, and nothing disagreed. It now renders on /about, after the MindHYVE section that already discusses the sister products, listing ArthurAI, ChironAI, JustineAI and EliAI with their reasoning-mode signatures and links to their grids. Nothing about Theo itself changed. Separately and in the same week, EliAI — the Agentic Finance Operating System, whose Digital Employee is Eli — joined the shared portfolio canon that all four grid sites carry verbatim, so the statement now reads five Operating Systems across five regulated verticals rather than four. The canon directory is synced by copy across the four repositories and the change was landed in all four together, because each repository verifies its canon only against its own manifest and nothing in the estate detects one site drifting from the others.

See the portfolio band →

Session replay removed from the chat, the console and the corpus site

What you ask Theo is a statement about your religious belief, and it is now recorded nowhere that can replay it. Microsoft Clarity — which records the page itself, not just events — has been removed outright from chat.theogrid.ai, from the internal console our team uses to review conversations, and from ipsc.theogrid.ai. Not masked or narrowed: removed, so there is no configuration left to get wrong. It remains only on these public marketing pages, where there is no conversation to record and consent is still asked first. Three specific leaks closed with it. The chat transcript was ordinary text on the page, so a replay captured every question and every answer, and the module claimed masking that only ever applied to form fields. A question shared as a ?q= link reached the advertising pixel, which reads the address bar in the page head before the application has started — that link is now stripped before any tag can see it. And free-typed help and glossary searches, which are frequently the question itself typed into the wrong box, were being sent with the words attached; the events remain, the words do not. The privacy policy has been rewritten to say what is now true, including where replay still runs.

What is and is not recorded →

Sanad, published as a specification before it ships

TheoAI Sanad now has a page, and it is the canonical statement of what Sanad is: give it an attributed claim — hadith or Qurʾanic — and it locates the claim in the corpus, walks the chain of transmission, reports the classical grade, and shows the evidence before it shows a verdict. Sanad has not shipped. The page is written throughout in a tense that says so, every capability on it carries an explicit status, and the only call to action is the waitlist — the same treatment /majlis and /mizan already get. Two things are deliberately absent: a narrator count, because at the time of writing the published figure did not reconcile with the index behind it (see the entry below), and a launch date, because none is set. One promise was taken back out before a reader could rely on it: there is no scholar-review lane, so the sixth verdict is “Unresolved” — an honest terminus that shows what was found and what would settle it — rather than a queue nobody staffs, and a contested verdict is published alongside its case rather than arbitrated. Sanad reports attribution and evidence. It does not issue rulings.

Read the Sanad specification →

The Evidence-Based Opinion boundary moved to the front door

“Evidence-Based Opinion” appeared on ten pages of this site and on neither of the two that a reader arriving from social actually lands on. The home page and /about now both carry the boundary, in the wording already approved on /theo/ and already present in all five languages — reused rather than re-authored, because the four non-English renderings of a religious-boundary claim are precisely the ones we are least able to review ourselves. The /theo/ page was also contradicting itself: one capability card promised scholarly attribution for every ruling, on a page that elsewhere states plainly that Theo never issues one, and the four translations of that clause carried the term with less ambiguity than the English did. Corrected in all five. Theo is a scholar who returns an Evidence-Based Opinion with its chain of transmission and hands the decision back to the reader; that distinction is architectural, not cosmetic, which is why it now sits on the pages that receive the traffic rather than one click in.

What Theo does, and what he refuses to do →

The site declares its own index; two figures gain their context

theogrid.ai had no robots.txt at all, and the conventional sitemap path returned the 404 page — the only one of the six MindHYVE properties in that state. Nothing was being blocked, but the sitemap was never declared, so every crawler had to guess that this site’s index lives at a non-conventional path. There is now a robots.txt that declares the index, and a redirect from the conventional path to it. There are no Disallow lines, deliberately: everything published here is meant to be read. Two figures also gained the context they needed to be read correctly. The v4.0.0 tiles on /corpus now state plainly that they are release deltas rather than current totals, and that the narrator unit changed after that release, so the older document total is not directly comparable with the total published above it. And the team page now shows six-month headcount growth beside the year-on-year figure, so the shorter and more recent trend is visible rather than inferred from the longer one.

Corpus figures and their provenance →

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.

Spanduk persetujuan cookie + Kontrol privasi

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.

Pembangunan ulang halaman Tentang untuk investor

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 diluncurkan secara publik

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