AI should not dull the human mind. It should sharpen it.

Vision दृष्टिः

In an age where a machine will answer anything the instant you ask, it is worth being clear about what that costs. An answer received is not the same as an answer understood — and a mind that stops deriving things is a mind getting weaker, not freer.

Geervanee has not stood apart from this AI era, and it is not left behind by it. We have brought our own inheritance into it: the discipline of Pāṇinian grammar, carried now by AI's speed, reach and patience. Traditional wisdom and modern computation working the same problem — and taking the learner further than either could alone.

AI शिक्षकः is your Sanskrit tutor. It does not hand you the śloka's split, the form, or the translation. It guides you — a nudge, then a narrower hint, then the operative rule, and only at the very last resort the answer itself — by which point you have done the work. Where ancient wisdom meets modern AI, the aim is not a faster answer. It is a learner who can derive the next one, and the one after that, alone.

How it works रीतिः

The design is neuro-symbolic. A deterministic Pāṇinian engine (vidyut-prakriya, an implementation of the Aṣṭādhyāyī) decides what is true — the correct form, and the exact sūtra that produces it. The language model never decides grammar. It only phrases the hint the engine has already justified.

The withheld answer is not in the model's prompt until you have reached the last rung of the ladder. A hint can cite एचोऽयवायावः (6.1.78) because the engine walked that derivation, not because the model recalled a pattern. When the tutor is finally forced to reveal a form, it shows the derivation with it — so even the reveal is a lesson.

The suite साधनानि

Four tools, one Geervanee login. शिक्षकः is the home surface.

Roadmap आगामि

Where this is going. Dates are intent, not promises.

1. Now — live शिक्षकः across the core grammar.

   sandhi · vibhakti / vachana · śabda-rūpa · dhātu-rūpa · kṛdanta · taddhitānta · sanādyanta · kośa lookup — each hint verified and sūtra-cited.

2. Next The full pañcavidhā vṛtti.

   samāsa and ekaśeṣa verifiers, plus real sandhi-vigraha, added through Prof. Amba Kulkarni's Samsādhanī engine — so compound analysis is as sound as the rest.

3. In progress संशोधकः — deeper sentence checking.

   Speak or type your own Sanskrit sentence and it coaches you to fix it, hint-only — live in beta now (verb, doer, agreement, word validity). Full kāraka / word-order checking lands with the Samsādhanī engine above.

4. Later वाचकः and लेखकः.

   Metrical chant reading, and recitation turned into a verified, exportable transcript.

5. Ongoing A Geervanee open model.

   Every verified hint the tutor gives is kept, with consent, as training data for a small open model of our own — so the tutor stays cheap enough to keep free.

6. Everywhere One tutor across every Geervanee platform.

   The same AI शिक्षकः, the same verified grammar, on geervanee.in, geervanee.com, the Samsad app, the Geervanee mobile app, and Kshanasah — one Geervanee ID for all of them.

Who builds this

Geervanee Educational Services, Chennai — the same people behind geervanee.in, geervanee.com, the Samsad app, the Geervanee mobile app, and Kshanasah. AI शिक्षकः is the tutor across all of them.

Questions, corrections, or a grammar case the tutor gets wrong — [email protected]. Reports of a wrong hint are the ones we act on fastest.