Indian AI research lab
Kalpna
Where imagination
takes form.
कल्पना से आकार तक
India's AI research lab building foundational models for language, vision, speech and documents.
- Voice AISpeech IntelligenceSpeech-to-text and text-to-speech for Indian languages.
- Vision-LanguageVisual UnderstandingModels that see a scene and describe it.
- EulerFlowAI WorkflowsNo-code builder that turns models into working services.
- OCRDocument IntelligenceReads print, handwriting and old scripts.
- SLMsLanguage ModelsCompact language models that run close to the data.
- Research-Grade Models
- No-Code AI Workflows
- Ready Assistants for India
- Build · Deploy · Scale
Our belief
Intelligence rootedin imagination.
We combine mathematical thinking, Indian knowledge traditions and modern AI research to create intelligent systems for the world.
India drew algorithms in rice flour long before it wrote them in code. We build India's models for the world — in that same unbroken line.
Our models
A family of foundational models.
Five model families, one shared vision — building intelligent systems grounded in India's languages, knowledge and imagination.
Voice AI
Speech intelligence for Indian languages.
Speech-to-text · Text-to-speech · Voice agents
Small Language Models
Efficient intelligence for Indian languages and domains.
Compact models · Domain adaptation · Efficient inference
Vision-Language
Models that understand what they see.
Image understanding · Video · Visual question answering
OCR
Reading the world's written knowledge.
Printed text · Handwriting · Historical scripts · Documents
Multimodal AI
Understanding across multiple modalities.
Text · Vision · Speech · Cross-modal reasoning
EulerFlow
Turn ideas intointelligent workflows.
A no-code platform to build, run and deploy AI pipelines — from RAG and vision to geospatial analysis and robotics.

EulerFlow
- Read DocumentsIngest PDFs, scans and email; extract text, tables and fields.
- Understand with AILanguage models, RAG and reasoning over what was read.
- Analyze (Vision / Geo)Vision models and geospatial analysis on images and maps.
- Act (Notify / Approve)Send alerts, file records, or wait for a person to approve.
- 300+ready blocks
- Visual, no-code canvas
- Test · Deploy · Scale
Model experiences
Explore what ourmodels can do.
A first look at the capabilities we're bringing together.
Voice AI
Phase 3Speech intelligence for Indian languages.
Preview (Hindi)“भारत की प्राचीन ज्ञान परंपरा आज भी हमें नई खोज के लिए प्रेरित करती है।”
OCR
Coming soonReading and understanding written knowledge.
Preview (Extracted Text)विद्या ददाति विनयं विनयाद् याति पात्रताम्। पात्रत्वात् धनमाप्नोति धनात् धर्मं ततः सुखम्॥
Vision-Language
Phase 3Understanding images with natural language.
Preview (Caption)“Two deer standing near a forest hut.”
Developers
Illustrative · API not publicBuilt for Indianscripts, end to end.
Devanagari in, Devanagari out. Model specs and pricing will sit here once the API opens.
curl $KALPNA_API_URL/v1/translate \ -H "Authorization: Bearer $KALPNA_API_KEY" \ -d '{ "text": "कल्पना से आकार तक", "source": "hi", "target": "en" }'



