SPRAYENA ADVISORS PRIVATE LIMITED

In their own words

What we're building: Nikitria is a local-first desktop app where a user assembles a personal "panel of experts" out of small, specialized language models that run entirely on their own machine. When they ask a question, a lightweight router decides which experts are relevant, sends the query out, and a synthesis layer composes the answers into one refined response, with a provenance bar showing exactly which model said what, and any disagreement between experts surfaced openly instead of blended away. The long-term business isn't the app itself: it's a certified marketplace where creators sell specialized SLMs, every listing passes our own adversarial test pipeline, and we take a revenue share. The marketplace and its two-sided network effects are the actual company.How we're helping: Three real problems get solved at once. First, privacy and ownership - everything runs on the user's hardware, offline if they want, with no data leaving the device. Second, trust - today's AI answers are opaque blends; ours show attribution, confidence, and contradiction, so the user sees a transparent panel of experts rather than a hidden blender. Third, creator monetization - the GPT Store proved demand for creator-built AI but paid creators pocket change; we give specialist model-builders a real, certified storefront with creator-friendly revenue splits.Why it's the future: The industry has already proven the three pillars this rests on. Small fine-tuned specialists beat frontier generalists on their own turf (fine-tuned Gemma, BioBERT, LinkedIn's EON-8B). Combining models can beat any single model, Mixture-of-Agents topped GPT-4 Omni on AlpacaEval using only open-source models. And local inference has gone mainstream, with 3B–8B models now delivering quality that needed 30B+ two years ago. Routers and runtimes are commoditizing fast; the durable moat is the curated, certified network of specialist models and the trust layer around it. Frontier AI is centralizing intelligence in a few clouds. we're betting the counter-trend: many small, honest, owned experts on your own machine, and a marketplace that makes building them a living.

Written by the company on its public Startup India profile. Not verified by us.

Is this your company?

Everything here comes from public government records. If something is wrong, the authoritative place to correct it is the Startup India portal itself, and we will pick the correction up on the next monthly snapshot. To flag an error on our side, use the report link or write to trust@indianstartupmap.com.

How we built this record Where the data comes from How you may reuse it