Your documents, turned into answers you can prove.

Not a sample. Every one of the six documents Inoapps supplied is read and represented, each atom attributable to the document and section it came from. This is the coverage, shown in full.

Pick a domain, ask in plain English. Switch tabs to see not just the answer but how it was reached, what it depends on, and what it can’t answer. Every reply carries a seal you can break.

Thea isn’t just ours to fill. Paste your own document, name its source, and it becomes a sourced, sealed corpus you can question in seconds — no code, no hand-authoring. This is the same engine a developer calls with three lines.

The 18-step setup isn’t a list — it’s an order with real prerequisites. This graph is built from the functional process doc’s own stated dependencies. Ask “traverse” in the console above to walk any step back to its roots.

The same architecture, applied to maths. Type an idea; it compiles to a glyph program, runs on real deterministic engines, and seals the result. Identical idea → identical hash, every run — the property an LLM cannot have.

One import. The demo above calls this exact file. Every method is real, tested, and runnable right here.

// the whole SDK — no build step
import { Thea, verifyChain } from './thea.js';
import { Glyph } from './glyph.js';

const thea = new Thea({ supabaseUrl, supabaseKey });

// one call — decided, sourced, sealed
const r = await thea.query(q, { vertical });
r.answer; r.path; r.sources; r.receipt;

// inspect & traverse
await thea.explain(q);     // why this answer
await thea.traverse(q);    // prerequisite chain
await thea.coverage(probes);// gap map
await thea.queryAll(q);    // cross-domain

// throughput & UX
await thea.batch(items);    // many, one chain
await thea.stream(q,{onToken});
thea.on('query', fn);   // events

// GlyphMath — real maths, sealed
await new Glyph().reason(idea);
verifyChain(thea.receipts());