The biggest AI story this week did not come out of a San Francisco lab. It came out of New Delhi. On June 12, an Indian startup shipped a homegrown video model. It generates clips at a fraction of what Western tools charge. So this is the India sovereign AI video moment people kept predicting, and it landed quietly. Meanwhile, a top US fund put $35 million behind software that runs small businesses on their own. And HR teams woke up to a hiring problem that the same AI boom created. Three threads, one question for your team: where does cheap, capable AI help you, and where does it quietly cost you?

India’s Sovereign AI Video Model Arrives Under the IndiaAI Mission

Avataar.ai launched Varya on June 12 at an event in New Delhi. MeitY Secretary S. Krishnan attended. The company calls it India’s first distilled video model. It was built after the IndiaAI Mission selected Avataar to develop indigenous foundation technology on subsidised national compute. (Source: Analytics India Magazine)

The numbers are the point. Varya is a 14-billion-parameter model. It uses distillation to cut the generation process from 50 steps to four. Avataar says output quality matches Alibaba’s Wan 2.2 and Google’s Veo 3. But it runs at roughly ₹0.48 per second. That is up to 10 times cheaper than leading global models, according to the company. (Source: Republic World)

Why India’s Sovereign AI Video Push Matters for Your Team

For founders and HR leaders, this is not about marketing clips. It is about cost and control. Say you run a distributed team across India. Cheap video generation then means training, onboarding, and recruitment content in regional languages, without a studio budget. For example, a model built for India’s languages handles Tamil, Bengali, or Marathi internal comms in a way foreign tools rarely do well. So the India sovereign AI video story is really a story about access. Who can afford to produce content at scale, and in which languages?

There is a compliance angle too. A sovereign model trained on national infrastructure keeps sensitive workforce content closer to home. That matters as India’s data rules tighten. What to do this week: pick one use case. Maybe a multilingual onboarding video, or a role explainer for hiring. Then test whether a cheaper Indian model clears your quality bar before you renew a pricier Western contract.

a16z Bets $35M on AI That Runs Small Businesses

Lassie raised a $35 million Series A led by Andreessen Horowitz. That takes total funding to $47 million. The founders came from Robinhood, Coinbase, and Superhuman. Their startup builds AI that runs back-office operations for small firms. Specifically, it automates claims, payments, and revenue reconciliation, rather than just suggesting actions. (Source: a16z)

Lassie says it already operates inside more than 700 businesses across 49 states. It claims over 250,000 hours of labour saved a year. So what does this mean for you? For a lean ops or HR team, this is the shape of the next wave. It is AI that does the work, not AI that drafts a memo about the work. Still, the risk sits where it always does. When software acts on payroll or payments on its own, your review process becomes the last line of defence. So build that in before you switch anything on.

Skillfishing: The Hiring Problem AI Just Made Worse

As AI helps recruiters move faster, it also helps candidates fake the skills you are testing for. The trend now has a name: skillfishing. For instance, invisible overlay tools can feed AI answers onto a candidate’s screen during a video interview. The interviewer sees nothing. According to Skillsoft, 91% of HR leaders believe employees overstate their skill proficiency, especially in AI, leadership, and technical work. (Source: Skillsoft)

For any team hiring in 2026, this changes how you screen. The advantage no longer goes to whoever hires fastest. Instead, it goes to whoever verifies earliest. (Source: HR Dive) So move work samples and structured, real-task assessments to the front of your funnel. Because if an AI overlay can pass your interview, you are not testing the candidate. You are testing the tool. Many teams already use AI in recruitment to speed sourcing. The fix is to spend that saved time on deeper verification.

A New AI Workforce Platform Wants to Retire the HRIS Mindset

On June 12, HR technology firm Tapplent launched iHVI, short for Integrated Human Value Intelligence. It pitches one platform for recruitment, onboarding, planning, learning, and management. Moreover, it is built around agentic workflows and conversational interfaces, not static records. (Source: India Press Release)

Vendor language aside, the direction is real. HR software is shifting from systems that store data to systems that act on it. For a growing company, the practical test is simple. Does your current stack expose clean APIs that AI agents can read and write to? If not, that integration project comes first. Any agentic feature comes after. So it is worth comparing what you have against the AI agent workflows these platforms now promise.

Quick Hits

• Apoha exits stealth with $36M. The London and San Francisco startup raised a Series A led by Singular on June 3. It applies its “Liquid State Intelligence” to predict how molecules and materials behave for pharma and battery research. (Source: Fortune)
• India and Nepal open AI ties. The two countries held a June 12 seminar on AI collaboration. Meanwhile, a second cohort of 25 Nepali startups began an eight-week, fully funded AI program at IIT Madras Pravartak. (Source: The Tribune)

The thread tying today together is access. Cheaper models, AI that does real work, and platforms that act on their own all lower the cost of capability. So if you are building a team in India or hiring across borders, Asanify’s HRMS keeps your people data structured and API-ready. That is exactly what these new AI tools need to plug into. For founders scaling in the region, our guide on how to hire AI engineers in India is a good next read.

FAQ: India’s Sovereign AI Video and the Week in AI

What is India’s sovereign AI video model?

It is Varya, a video-generation model from Avataar.ai launched on June 12 under the IndiaAI Mission. The 14-billion-parameter model uses distillation to generate video in four steps instead of 50. Avataar says it runs up to 10 times cheaper than leading global tools, while supporting India’s languages.

Can AI really run back-office work for a small business?

Increasingly, yes, for narrow tasks. Lassie, backed by a $35 million round from Andreessen Horowitz, automates claims, payments, and reconciliation across more than 700 small businesses. Still, the catch is oversight. When AI acts on money or payroll directly, your review and approval steps become the safeguard. So keep a human in the loop.

How do you stop skillfishing in AI-era hiring?

Move verification earlier in the process. Skillfishing happens when candidates use AI tools to fake skills in resumes and live interviews. The fix is work samples and structured, real-task assessments at the front of the funnel. It also helps to use interview formats that are hard for an on-screen AI overlay to pass.

Not to be considered as tax, legal, financial or HR advice. Regulations change over time so please consult a lawyer, accountant  or Labour Law  expert for specific guidance.