Hashtee logo
Hashtee

AI for the real world.

Models and applications that hear, see, read, and act — in 30+ languages, running in production today.

Built for production. Deployed worldwide.

Every language, not just English.

Our speech and vision models are trained from the ground up for 30+ languages — Hindi, Tamil, Arabic, Japanese, and more. Not English models with a translation layer.

Running in factories today.

Every model ships after validation against real production data. Obi monitors assembly lines. Dodo inspects parts. Kundoo transcribes call centers. This is not a demo.

Your infrastructure, your rules.

Deploy on-premise, in your private cloud, on edge devices, or consume as APIs. We fit into your setup — not the other way around.

Engineers, not just a dashboard.

We work with your team from model selection through deployment. When something breaks at 2am on a production line, we're in the same Slack channel.

Running in production.

All stories
Grasim Industries automates yarn cone inspection and doff card digitization with Raru and Longlichi.
Hashtee logo
Hashtee
Textiles & Viscose
Schneider Electric India analyses internal documents and competitor intelligence with custom RAG pipelines.
Hashtee logo
Hashtee
Energy & Automation
Raru Runtime and Binbin inspect fasteners and track WIP across Metaforge's production floor.
Hashtee logo
Hashtee
Fastener Manufacturing

What you can build with Hashtee.

Voice that gets every language right.

Transcribe and generate speech across 30+ languages — including Hindi, Tamil, Telugu, Arabic, Japanese, and more. Handle code-mixed conversations that other models get wrong.

Kundoo STT — real-time transcription
Shama TTS — natural multilingual voices
Voice Agent — production-ready in minutes
KundooSpeech-to-Text
Hindi + English

Transcript

Haan sir, aapka order #4821 kal dispatch ho gaya hai. BlueDart se delivery hogi, expected date hai 9 April. Kya aap tracking number chahte hain?

Inspection without inspectors.

Detect surface defects, measure dimensions, and make pass/fail decisions faster than any human QA team.

Powered by Raru Runtime
Sub-50ms per-part inspection
Improves with every production run
PASS
48.2mm
DEFECT
scratch
SCANNING
Live
Camera 338ms
Raru Runtime v2.1

Recent inspections

#4823Pass
2s ago
#4822Reject — surface scratch
5s ago
#4821Pass
8s ago

Machines talk. Obi listens.

Every machine has an acoustic signature. Obi learns what healthy sounds like — and alerts you before a bearing fails or a motor seizes.

Anomaly detection from sound
Works with any microphone
Weeks of early warning
Monitoring — Line 3
Normal
72 dB·Motor hum·2s ago
Anomaly
94 dB·Impact detected·Just now

Paperwork at machine speed.

Feed Longlichi a photo of a receipt, dispatch note, or handwritten tag. Get clean structured data back — in any language, any script.

Receipts, invoices, tags, plates
Handwriting and print
JSON output for your ERP
InputDispatch note DN-0472
Output200 OK — 340ms
{
"doc_type": "dispatch_note",
"number": "DN-0472",
"date": "2026-04-07",
"copy": "customer",
"party": "RaC Industries",
"items": [{ "name": "Bearing Assembly", "qty": 240 }],
"weight": "1200 kg",
"lorry_no": "TN 09 AB 4821",
"signed_by": "राजेश कुमार"
}

Customer support that never sleeps.

Pechi handles order inquiries, reschedules deliveries, and resolves complaints — on WhatsApp, phone, and web. Your team handles the exceptions.

Resolves, not just responds
WhatsApp, voice, web, API
Hands off to humans only when needed
P
Pechi Assistant

What's the status of order #4821?

Order #4821 shipped yesterday via BlueDart. Expected delivery: April 9. Tracking: BD2847291.

Can you reschedule it?

Done — rescheduled to April 11. Updated SMS sent to the customer.

One platform. Every modality.

Apps

Pechi
Dodo
Binbin

Platform APIs

Speech-to-Text
Text-to-Speech
Voice Agent
Acoustic
Vision
Document

Models

Kundoo
Shama
Longlichi
Obi
Raru Runtime

Infrastructure

On-premise
Private Cloud
Edge Devices
Hashtee Cloud

Services

Custom ModelsYour data, your model
IntegrationERP, MES, SCADA
DeploymentWe handle rollout

AI deployments designed for privacy.

Deploy our models in your environment — on-premise, edge devices, or cloud. Your data never leaves your premises.

Discover more
AWS
NVIDIA
Docker
Raspberry Pi
ONNX
Jetson
Linux

We're hiring.

Engineers who build AI that ships — not AI that demos.

View open roles