No books to buy. The Greek, the Hebrew, the parsing, the lexicons, the grammar, the manuscripts, the narration — it's all in, for everyone, and no book behind it has a price tag. You don't own a shelf; you open the Bible.
We tell you what the text says, not what somebody thinks about it. A library is other people's conclusions, and an AI over a library hands you the loudest one. Ours reads the words — every claim tied to a form you can tap, every reference checked, and where the text is silent, we say so. Commentators are there when you ask for them, named and dated, as opinion. The Bible is the source. Everything else is a footnote.
Free original-language engine forever · upgrade only when you want the AI to do the heavy lifting.
The original-language text with every word parsed, the lexicons, the grammar rules by name, the manuscript apparatus, a cross-testament bridge to the Septuagint, our own formal translation and its narration — all of it is in from the first minute, free, with nothing behind a book price. The commentaries you already own stay where you own them: on your shelf, your Kindle, your other app. We read the text; you keep your library.
“Nothing to buy before you can read a word.”Every answer starts from the words on the page — a form you can tap, a rule by name, a reference we checked against the text before you saw it. Where a claim is an interpretation it is labelled one; where the text is silent, the answer says so instead of filling the gap. The commentators are a tap away when you want them, named and dated, as opinion beside the text — never as the text. The Bible is the source. Everything else is a footnote.
“Don't trust it. Check it — every claim shows its source.”And when you have found something: save it, shape it into a study or a sermon, take it with you. Your notes are yours, in a form you can download whole, and leaving is never expensive.
I build AI systems for a living at a Fortune 500 company. I also spent four years at a Bible college in the Midwest, minored in Greek, and served as a youth pastor for thirty years.
That combination is the whole reason this exists. Building these systems professionally means I know exactly how they fail — fluently, confidently, and most dangerously when they are almost right. Thirty years in ministry means I know what it costs when someone stands up on a Sunday and says the Greek means something it never meant.
Most people rightly worried about AI and scripture can't audit the machine. Most people building AI for scripture have never had to answer for what they taught. I've spent my life on both sides of that line.
The danger was never that a computer touches the text. It's unearned confidence — a plausible sentence about a word you have no way to check, delivered in a voice that sounds like scholarship. I have watched general-purpose models invent parsings, assign words meanings they never carried, and cite verses that say nothing of the kind. In most fields that's an irritation. Here it becomes a sermon, and then it becomes what a room full of people believe.
So my job on this project is not to make the AI smarter. It is to keep it honest: tethered to the actual text, and made to show its work — every claim carrying where it came from, so you can go check it yourself.
That is why answers cite their sources, why each claim is labeled with its provenance, why a word note says UNSURE rather than guess, and why anything the model recalls from memory instead of reading from the corpus is marked as exactly that. The standard I'm building to is simple, and it is not "trust me."
“You should never have to take my word for it — or the machine's. You should be able to check it.”
Scriptura is an MCP server — the open standard for giving AI agents tools. Point Claude Code, Claude Desktop, or any MCP-capable harness at your key and your own model gets 30+ research tools over the full tagged Greek & Hebrew corpus: passages with morphology, word studies, a grammatical concordance, construct search (find a form within N words of another — the query power users pay hundreds for elsewhere), textual variants with the apparatus, lexicons, Calvin's Institutes, and more. Your model does the reasoning on your subscription; our corpus does the grounding. No scraping, no PDFs, no copy-paste.
claude mcp add scriptura --transport http https://YOUR-DOMAIN/mcp \ --header "Authorization: Bearer YOUR-KEY"
If you live in the languages, this is where you should end up: your own Claude, researching the real corpus inside your own workflow. Included with Starter, Standard and Pro, or on its own as the Connect tier under pricing below. Full setup guide →
Subscribe and your MCP key is emailed to you automatically — no ticket, no wait. Building a product on the corpus (not just your own study)? Commercial redistribution tiers exist — write us at hello@biblelabs.ai and we'll set you up.
Paid plans open shortly. Everything free is free today — sign in and read, listen, mark and write. The plans below are what is coming and what they will cost; you'll be able to buy one here the day billing opens.
For scale: a traditional study-software library runs $500–$2,000 up front, before you've bought a single commentary. Everything below costs less per year than one good commentary set.
Choose this if you want to read the original languages and study by hand — no AI needed.
Choose this if you want the app to do the heavy lifting — grounded answers and drafts, no setup.
Choose this if Starter's 300 questions run out before the month does, but you're not in the text every day.
Choose this if you're in the text every day and keep hitting the Starter limits — volume is the point.
Annual is one month free, paid up front. Your allowance is still monthly — 600 questions a month on Standard, not 7,200 for the year — and unused questions don’t roll over into the next month. Cancelling an annual plan refunds the unused whole months. Top-up credits are the exception: bought outright, they never expire.
The software doesn't do your theology — you do. Magnify just amplifies your strength and speeds your output, grounded in the languages, at a price that respects your calling and your budget.
Sign-ups open soon — the corpus, the tooling, and the MCP server are already running.