Korvo Intelligence Engine

Meet Medha.

मेधा - Sanskrit for intelligence

On-device intelligence that verifies what AI models produce. Consensus synthesis, contradiction detection, experiment evaluation - under 10ms, zero tokens, completely offline.

Your models generate. Medha verifies. You decide.

Or upgrade to Pro - $149/year →

Medha ships bundled with Korvo · API & HuggingFace coming soon

The Problem

AI models generate confidently.
Nobody checks their work.

You send a question to GPT-4, Claude, and Gemini. You get three plausible, confident answers. Which claims are actually agreed upon? Where do they contradict? What's uncertain? Today, you either read all three manually - or send them to yet another cloud model for synthesis, paying more tokens and waiting more seconds.

Cloud synthesis is slow and expensive

Sending multiple AI drafts to another LLM for evaluation costs tokens, takes 5-10 seconds per call, and requires internet. At scale, it's unusable.

No structured disagreement analysis

Models give you text. They don't give you "Claim A from GPT-4 contradicts Claim B from Claude with 73% similarity." You're left reading and comparing manually.

Evaluation at scale is a cost wall

Running 100+ experiment iterations overnight? At $0.03-0.10 per evaluation, that's $3-10 per run. Most people just... don't run experiments.

Architecture

Where Medha sits in your workflow.

Medha doesn't replace your AI models. It sits between them and you - verifying, scoring, and synthesizing before anything reaches your decision.

Your AI Models (BYOK)

OpenAIAnthropicGeminiOllama

They generate text, answer questions, create drafts

↓

Medha (on-device · Rust · <10ms · $0)

Claim extractionConflict detectionConsensus synthesisEval scoring

↓

You

Decisions backed by verified, measured, traceable evidence

Capabilities

What Medha does - specifically.

Consensus Synthesis

Shipped

When you run Multi-Agent Consensus in Korvo, multiple AI models answer the same question in parallel. Medha synthesizes the results locally - no cloud round-trip, no extra tokens.

Claim extraction from each AI response

TF-IDF vectorization for numerical comparison

Graph clustering to find agreement across sources

Conflict detection with similarity scoring

Authority scoring with hedging-language detection

Final synthesis with confidence score (0.0–1.0) and reasoning chain

Contradiction Detection

Shipped

Medha identifies where AI models disagree - not just surface-level differences, but actual conflicting claims backed by evidence from different sources.

Pairwise claim comparison across all agent drafts

Cosine similarity scoring to surface real conflicts

Ranked disagreement map with source attribution

Uncertainty surfacing - know where models are guessing

Experiment Evaluation

Building

Korvo's AutoPilot runs experiment loops overnight - modify, measure, keep or discard, repeat. Medha makes the "measure" step instant and free.

medha_eval_claims - precision, recall, F1 against gold-standard claims

medha_eval_retrieval - nDCG, MRR, Precision@K, Recall@K

medha_eval_coverage - reference claim coverage via sliding-window TF-IDF

medha_suggest_params - next-experiment suggestions via explore/exploit balancing

The Math

The difference between 5 experiments
and 500 experiments.

When Korvo's AutoPilot runs experiment iterations overnight, the evaluation step determines how far you can go.

Cloud LLM Eval

Medha (on-device)

Cost per eval

~$0.03–0.10

Included with Korvo

Time per eval

3–10 seconds

<10 milliseconds

100 iterations

$3–10 · 5–17 min

$0 · 1 second

Internet required

Yes

Works offline

Yes

Data leaves device

Yes

Never

This is the difference between “I ran 5 experiments because tokens are expensive” and “I ran 500 experiments while I slept.”

Technical Specs

Medha v1 - what ships today.

Native Rust library, compiled per platform, integrated via Dart FFI.

Language

Rust (native, compiled per platform)

Platforms

macOS, Windows, Linux, iOS, Android

Integration

Dart FFI - C function interface from Flutter

Binary size

~2 MB

Average latency

<10ms per operation

Dependencies

Self-contained - no runtime, no Python, no JVM

Network calls

Zero. Medha never touches the internet.

GPU

Not required - runs on CPU

Cost

Included with Korvo. No per-token billing.

Coming Soon

Roadmap

Medha beyond Korvo.

Medha ships bundled inside Korvo today. Soon, we're bringing it to developers and researchers as a standalone capability.

Medha API

Call Medha's consensus synthesis, claim evaluation, and retrieval scoring from your own applications. REST API with sub-100ms response times.

Consensus synthesis endpoint

Claim extraction & comparison

Eval scoring (nDCG, MRR, F1)

Batch evaluation for pipelines

POST /v1/consensus/synthesize
POST /v1/eval/claims
POST /v1/eval/retrieval

HuggingFace

We're training and refining Medha's evaluation models and plan to publish weights on HuggingFace - so researchers can run, fine-tune, and benchmark independently.

Pre-trained evaluation weights

Claim extraction model

Consensus scoring model

Research-friendly licensing

🤗 korvo/medha-eval-v1
🤗 korvo/medha-consensus-v1
🤗 korvo/medha-claims-v1

Interested in early API access or research collaboration? Get in touch

Honest Positioning

What Medha is - and isn't - today.

✓A deterministic reasoning engine for verification and evaluation

✕A chatbot or text generator (today)

✓Complementary to OpenAI, Anthropic, Gemini - it verifies what they produce

✕A replacement for your AI models

✓Required: zero. Korvo works perfectly without it via cloud synthesis

✕A dependency - it's an accelerator

✓Evolving - deeper reasoning capabilities on the roadmap

✕Staying still - Medha gets smarter with every Korvo release

FAQ

Questions we get asked.

Is Medha an LLM?

Today, Medha is a deterministic reasoning engine - claim extraction, consensus synthesis, evaluation scoring. It doesn't generate text or hold conversations. Its capabilities will expand with every release. The core promise stays the same: intelligence that runs on your machine, not someone else's server.

Why not just use GPT or Claude for synthesis?

You can - Korvo lets you toggle between cloud synthesis and Medha synthesis. The difference: cloud synthesis costs tokens, takes 5-10 seconds, and requires internet. Medha does it in under 10ms, offline, for free. When AutoPilot runs 500 experiment iterations overnight, that difference is existential.

What if Medha isn't available on my platform?

Korvo falls back to cloud-based synthesis seamlessly. You won't notice a difference in the UI - just in speed and cost. Medha is an accelerator, not a requirement.

Does Medha send any data anywhere?

No. Medha is a compiled native library that runs entirely on your CPU. It makes zero network calls. Your data never leaves your device. This is an architectural guarantee, not a policy checkbox.

Will Medha be available outside of Korvo?

Yes. We're working on a standalone Medha API and plan to release weights on HuggingFace. Developers and researchers will be able to use Medha's evaluation and synthesis capabilities independently.

Is Medha open source?

Not yet. The core engine (libmedha_ffi) is proprietary and ships bundled with Korvo. We're evaluating open-sourcing components - particularly the eval functions - as Medha matures.

Will Medha ever generate text?

Medha's capabilities will expand. Today it verifies and evaluates. The roadmap includes deeper reasoning capabilities. We'll share more when we're ready.

Your models generate.
Medha verifies.
You decide.

On-device intelligence that never touches the internet. Ships with Korvo. API and HuggingFace coming soon.

Consensus synthesisContradiction detectionEval scoring<10ms latencyZero tokensFully offlineAPI coming soonHuggingFace coming soon