Trust & Verification

Accuracy is the whole brand. This page sets out — in plain terms and without marketing gloss — the standard every question has to clear before it reaches one of your students, what happens when a question fails it, and where the human stays in charge. These are commitments you can hold us to; the machinery behind them is ours, and it changes.

The Akriva Verification Engine

Every AI-drafted question passes through the Akriva Verification Engine before any human sees it. The engine runs on three cores: Psi builds the space of the paper, Nabla solves and stresses each question independently, and Lambda resolves the result. Where the cores disagree, the question is held back for faculty rather than shown to a student.

Faculty sign-off is the guarantee. The engine is what makes that sign-off fast. The AI does the tedious cross-checking so your teachers spend their judgement only where it matters — and nothing reaches a student until a human approves it.

What we guarantee

These are commitments, not a description of our internals — they are what you can hold us to, and they don't change when the engine does.

• Every question is solved independently of the process that wrote it, from scratch, without sight of the drafted answer.
• Where that independent work disagrees, the question is held back for your faculty — never quietly resolved and slipped into a paper.
• Questions are screened for the failures that actually cost students marks: ambiguity, missing information, and more than one defensible answer.
• A faculty member signs before any student sees it. This step is never skipped, and never automated away.
• No student data is sent to any AI provider, at any stage.

Which models we use

We use frontier models from multiple independent research labs, and we route each stage to the model best suited to it. We don't publish which model runs at which stage. That mix changes as capability and cost change, and how we combine them is our own work — it is the part of the engine we build, rather than license.

What we hold constant is the standard, and that is the part worth judging us on: every question is solved independently by models that did not draft it, disagreements are held back rather than resolved silently, and a faculty member signs before any student sees it.

What we do and don't claim

• We orchestrate, verify and prompt these models and own the question bank that compounds from them. We do not build or train the underlying foundation models — those are licensed.
• We don't claim a fixed “number of AIs” as a slogan; the standard is what matters, and the copy changes the day the pipeline changes.
• We don't publish superlatives we can't show. Prediction and accuracy numbers get published after results — not asserted before.
• We don't claim questions are flawless. Some are disputed, and those are the ones we hold back — that is the system working, not failing.

Your data stays yours

Verification runs on question content only. Student data never leaves Akriva and is never sent to any AI provider — which is the question people are really asking when they ask which models we use. Your question bank is private to your institute: we never share it with other institutes and never train models on it. All student data is stored in India. More detail is on our Privacy page.

Questions?

If you're an institute weighing Akriva and want to go deeper on any of this, hello@akriva.ai reaches a founder directly.