Vardhan Dharnidharka

AI/SaaS/Deeptech Investor @Stellaris_VP
ex-ML/Eng @Lyft, @ClickUp, @Bloomberg

[Vardhan Dharnidharka @Vardhan1119]

In the new world PMs work directly with customers and decide what to build. FDEs make it work in the real world. Everything in the middle is commodity

The new PM is not the old PM: part consultative seller, part co-builder, part customer success

At the early stage, that's the founder

[Lenny Rachitsky @lennysan]

STATE OF THE TECH JOB MARKET IN 2026

Twice a year, I dig into trends in hiring and job openings. This is the fifth report I've done, and the most surprising yet:

1. PMs are winning in the AI era
2. Engineering demand has stalled
3. Design openings are at a record low
4. Every role now requires AI skills
5. Tech jobs are concentrating in the Bay Area
6. Remote roles are at a record low
7. The tech job market is growing fast, but mostly outside of software (gasp!)

Full report here: https://t.co/Gly6uqwKxe

[Vardhan Dharnidharka @Vardhan1119]

Founders - deck or no-deck, and why? Fellow investors - seeing the same shift?

[Vardhan Dharnidharka @Vardhan1119]

Founders are preferring pitches with no deck - "let's just have a conversation." I have no problem with it, but it usually leads to a subpar outcome (actually I really like 1-pagers). Builders are poor natural storytellers, myself included & this is a time to play your strengths

[Vardhan Dharnidharka @Vardhan1119]

The middle ground: keep the story-telling, have the deck as a sidekick. Map, not script - retention question on slide 3? Jump to slide 11. An answer already sitting on a slide says you've thought through and a chance to show adjacent datapoints too

[Vardhan Dharnidharka @Vardhan1119]

This is a great outcome! We are quickly developing, deploying and competing against the latest and greatest. The culture is important

[Sarvam @SarvamAI]

Today we’re releasing Sarvam Prism, a System One model built on Sarvam 105B for fast, accurate decisions across enterprise workflows.

It scores 71.8 with a median response time of 44 milliseconds, the highest average score among decision models that respond in under 100 milliseconds.

[Vardhan Dharnidharka @Vardhan1119]

The thing I missed the most about AI/ML is becoming mainstream again https://t.co/bWNmEQ7s15

[Naman Jain @namanjn1998]

Your voice agent can get every word right and still get the call wrong.

Introducing Tazo (@TazoLabs), an audio decision model for live voice agents. Speech in, decision out.

Give it raw call audio, a text prompt and your options. It returns the calibrated probability for each option, directly from audio.

• 70.9% vs 46.4% for STT + Jev on real production calls
• Ranks #2 on audio intelligence benchmark (MMAU)
• <200 ms model latency
• First decision model tuned to run on streaming voice calls
• Introducing “None of the above” option to handle out-of-scope and unsure states.

API is live. Free credits for every new sign-up.
Give your agent ears @ https://t.co/Czpr63FjlI

See it in action below 👇

[Vardhan Dharnidharka @Vardhan1119]

Is your AI managing your personal life this professionally?

[Vardhan Dharnidharka @Vardhan1119]

Big opportunity for the govt to be a market maker - absorb demand and fuel growth for the India-for-India usage via tax breaks and discounted pricing until we get to steady state in terms of utilization

[Vardhan Dharnidharka @Vardhan1119]

India has 2 categories of usage (mostly inference) - india for india and india for the world (majority driver given we are a Services-export economy)
Unfortunately, India for the world will require colocated data centers hence this capacity is of no use

[Chandra R. Srikanth @chandrarsrikant]

🚨 Imp: IndiaAI GPU providers may offer services to overseas clients to keep investments viable

The move follows concerns that demand for computing power in India is not yet sufficient to absorb the capacity being added under the government-backed programme.

Allowing companies to serve customers abroad would give them another source of revenue while the domestic market develops, a senior government official said.

"There is not enough uptake of GPUs in India right now. It is, therefore, important that these companies are able to sell the compute for the right price to anyone globally," the official said on condition of anonymity .

Under the Rs 10,372-crore IndiaAI Mission, the government has empanelled 14 companies to provide computing capacity using GPUs sourced from manufacturers including Nvidia, Intel and AMD.

The selected companies include Jio Platforms, Tata Communications and Yotta Data Services, alongside other data centre and cloud service providers.

By @AihikS

https://t.co/5KaETiykAx

[Alok Goyal @alokgoyal1971]

The one thing that separates a venture outcome from a good services business in AI native services is the slope. This is the piece we have spent most time on, because it is where we actually underwrite these businesses. https://t.co/ME7MkiwtYF

[Vardhan Dharnidharka @Vardhan1119]

@sanketnadhani Highly recommend!

[Vardhan Dharnidharka @Vardhan1119]

Jev will likely go down as one of the key inflection points in the AI super-cycle. I believe it will open up a whole new set of possibilities and will cannibalize a whole lot of tokens from the trad LLMs

[Vardhan Dharnidharka @Vardhan1119]

Moved a certain workflow from our @Stellaris_VP Brain to Jev - all inbounds (cold &amp; warm) are auto-triaged to the right team member (end CRM entry created)

I always felt trad LLM was a heavy hammer for this simple classification. Jev makes it a lot more natural (results below)

[Vardhan Dharnidharka @Vardhan1119]

We have taken vertical integration to a whole new level - start from application layer and raise for inference chips!

[Techmeme @Techmeme]

Source: AI assistant Instinct seeks to raise $1B at a $10B valuation, up from $2.25B less than a month ago; Sequoia and Benchmark are in talks to lead the round (@pauvalida / The Information)

(Visit Techmeme dot com for the link and full context!)

[Vardhan Dharnidharka @Vardhan1119]

4/ Big tech's problem: maximize RoI without blowing the budget. Startups' problem: one human running 20 agents without losing the plot

One is token-maxxing its way to PMF. The other is defending turf and justifying spend

Same tokens. Two problems. A company to build for each

[Vardhan Dharnidharka @Vardhan1119]

1/ One observation from my Bay Area trip last week: big tech and startups are still in different orbits — and now across a whole new dimension.

Whether they're token-maxxing. Or not. 🧵

[Vardhan Dharnidharka @Vardhan1119]

3/ Startups blew my mind. Each person runs 5-6 Claude / OpenAI Pro licenses at ~$200/mo each. They throw a Fable or Astra at everything. Velocity unlike anything I've seen. The only thing slowing them down: cognitive load of managing that many agents