TacnIQ.ai raises US$1.5 mil from In Group Holdings to scale tactile AI
- ID
- 27993
- Status
- summarized
- Published
- 23 Sep 2026, 4:11 PM
- Fetched
- 24 Sep 2026, 9:47 AM
- Provider
- Digital News Asia
- Category
- malaysia-tech
- Original URL
- https://www.digitalnewsasia.com/startups/tacniqai-raises-us15-mil-group-holdings-scale-tactile-ai
- Source URL
- https://www.digitalnewsasia.com/rss.xml
Summary
- Score
- 4.0
- Created
- 24 Sep 2026, 9:47 AM
- Tags
- Audience
- saas_foundersai_ml_learners
What happened
TacnIQ.ai has raised US$1.5 million (RM6 million) from lead investor In Group Holdings, the first close of a US$3 million (RM12 million) pre-seed round, to build tactile AI foundation models, hire engineers and scale deployments. The company says its platform, which lets machines interpret signals from physical contact, is already deployed with paying customers across logistics, construction, e-commerce, hospitality and healthcare, and that it has collected over 5,000 hours of tactile interaction data from controlled experiments plus commercial sites. In Group Holdings CEO Liu Song framed physical AI as a next frontier; TacnIQ.ai co-founder and CEO Aashish Mehta said the money goes to hiring engineers and turning the technology into scalable industry applications.
Why it matters
There is nothing here to adopt yet: no product, API, pricing, SDK or developer access is mentioned, so this is not a decision point for builders today. What is concrete is the go-to-market described in the text — putting sensor nodes into paying customers across five industries and using those sites as the data-collection layer for a foundation model, which is a template worth noting if you are pitching hardware-adjacent or robotics-adjacent AI in the region. The round is quoted in ringgit (RM12 million target) and sits in a Malaysia-tech category, so it is a data point on local capital appetite for physical AI, not a signal to change your stack.
Discussion angle
Is 'paying customers as sensor nodes' a repeatable playbook for physical-AI startups in Southeast Asia, or does owning a proprietary 5,000-hour tactile dataset only matter if the model generalises across those five industries — and what would you need to see to believe that claim?