Deepanshu Pathak

Building @getoasisstudio

[Deepanshu Pathak @Deepanshu016]

@itisRazak Hi 👋

[Deepanshu Pathak @Deepanshu016]

@JayBosshq Yeah, I can see the whole thing becoming much more contextual instead of tool-based.

[Deepanshu Pathak @Deepanshu016]

Creative software is going through a pretty interesting shift.
It used to give you tools and wait for you to make the decisions. Now it can understand intent, make some decisions, and move the work forward.
I’m curious how the interface changes when that becomes normal.

[Deepanshu Pathak @Deepanshu016]

Had a really good time here today.

Really enjoyed the talks and especially the hands-on workshop by @lee_stott , and I got to meet some really interesting people along the way.

Thanks for putting this together.
@hboelman
@GlobAICommunity

[Deepanshu Pathak @Deepanshu016]

Attending MCP Community Connect, Bengaluru
@GlobAICommunity

[Deepanshu Pathak @Deepanshu016]

Attending MCP Community Connect, Bengaluru
@GlobAICommunity

[Deepanshu Pathak @Deepanshu016]

Built this with @TigerGraphDB for HH Goa.
The interesting part was getting the agent to reason over connected evidence instead of treating a risk score as the answer.
Have a read: https://t.co/bWzAOAHBwO

@247pmstudio

[Gyanshu Pathak @gyanshu_pathak]

The future of motion designing is changing !

Recreated this ClickUp launch video with AI.

Took me just few minutes and 3 prompts in total.

Here’s the first 15 seconds. 👇

Waitlist is live - https://t.co/jTU4MZZ2e3

Want something like this for your product? DM me.

[Deepanshu Pathak @Deepanshu016]

Really excited to finally start showing people what we’ve been building. There’s a lot more coming with @getoasisstudio. Also the waitlist is live at https://t.co/pC4SmgQL58

[Gyanshu Pathak @gyanshu_pathak]

AI video generation is getting easy. AI with taste is the hard part.

For this one, I basically gave Oasis the idea I had in mind.

It figured out the visuals, generated the clips, and built the story around it.

I didn't have to describe every shot or write a detailed storyboard.

That's the direction I'm excited about:
AI that doesn't just generate. AI that understands what you’re trying to create.

This is how we are creating product videos at @getoasisstudio , waitlist is live now at https://t.co/CgA8EwArBO , if you want any videos like this please dm me !!

[Deepanshu Pathak @Deepanshu016]

Really excited about where we’re taking this. @getoasisstudio

[Gyanshu Pathak @gyanshu_pathak]

Motion designers are cooked. 👀

This is what making a product video can look like in the AI era.

The client gave me: → their website
→ the kind of video they wanted
I took it from there — story, visuals, motion, voiceover, everything.

No massive brief. No 20-page creative document.

Just a website + an idea → a high quality finished video in minutes

This is how we are creating product videos at @getoasisstudio , waitlist is live now at https://t.co/I6NBVDxpdy , if you want any videos like this please dm me !!

[Deepanshu Pathak @Deepanshu016]

Really interesting evening at the Unicorn AI Summit.

There’s a lot happening across the AI ecosystem right now, but what stood out to me was how differently people are approaching the same shift. Different problems, different products, different bets.

@PromptQL @grayscale_vc

[Deepanshu Pathak @Deepanshu016]

Unicorn AI Summit, Bengaluru

[Deepanshu Pathak @Deepanshu016]

Unicorn AI Summit, Bengaluru

[Deepanshu Pathak @Deepanshu016]

PUBLIC BUILDER at @EmberGround AI Hackathon

📍 Bangalore | Sep 27 | 120 Builders
💡 Building AI for India

Powered by @render @elevenlabsio @n8n_io @dodopayments @breeth

[Deepanshu Pathak @Deepanshu016]

Shipped a failure-injection harness against Nasiko’s A2A proxy.
Asserts 401 (no session / revoked jti), 429 on POST /api/orchestrator/a2a (30/60s), I8/I9 (dead and truncated upstream, /health 200), loopback bypass as ENVIRONMENT.

@nasikolabs @anakinHQ @DronaHQ

[Deepanshu Pathak @Deepanshu016]

At the Nasiko Build-A-Thon today.
Looking forward to a good day of building.
@nasikolabs @DronaHQ @anakinHQ

[Deepanshu Pathak @Deepanshu016]

Congratulations Gyanshu!

[Gyanshu Pathak @gyanshu_pathak]

Officially a Graduate now 😊🔥 !

[Deepanshu Pathak @Deepanshu016]

The harness is starting to feel almost as important as the model itself. This paper is a really interesting example of what happens when you actually optimize that layer.

[elvis @omarsar0]

Build your own harness, folks.

This is absolute banger paper from NVIDIA on self-evolving agent harnesses.

(bookmark it)

They introduce SoL-Pi which cuts token traffic by nearly half.

And it matches its baseline harness on GPT-5.6 Sol and Opus 5.

More details below:

Instead of tuning a harness by hand, they run auto-research loops at the harness layer across many repository-derived and verifier-driven environments, keeping only the mechanisms that survive selection.

Four mechanisms survived:

> Action Fusion changes how actions execute
> Online Context Compact handles compaction during a run
> ObservationPack reshapes observation handling
> Evidence-Preserving Reducer covers delegated reading

On the 51-task EdgeBench evaluation, the savings translate to about a third off API cost. In dollars that is an estimated $8.75 to $13.50 per hour against native Codex and Claude Code harnesses, and $4.36 to $5.71 against the baseline harness.

Because the search runs across many environments rather than one, the retained mechanisms keep working outside the setting that produced them. Code is on GitHub under NVlabs.

Paper: https://t.co/1x26LzuE6d

Chat with Paper: https://t.co/kygTc5XLFB

[Deepanshu Pathak @Deepanshu016]

I’m becoming more convinced that distribution should influence what you build, not just how you market it.

[Deepanshu Pathak @Deepanshu016]

Excited to be attending the Unicorn AI Summit on 21st Sep. If you are coming, let's catch up!

[Deepanshu Pathak @Deepanshu016]

A lot of thought goes into every frame here. This is the kind of quality we’re chasing with @getoasisstudio

[Gyanshu Pathak @gyanshu_pathak]

Motion designers are cooked now!

This is how we make product videos at @getoasisstudio , the first step is analyzing the user requirements and preparing a storyboard for the full video and this is the game changer step!

Checkout this one which I created for one of our client !

[Deepanshu Pathak @Deepanshu016]

This is crazy good....
Made completely with @getoasisstudio

[Gyanshu Pathak @gyanshu_pathak]

This will kill motion designers.

I made this 35-sec product video for a client.

No After Effects.
No video editor.
No traditional production workflow.
Just AI.

And this is only the beginning.

We're building Oasis @getoasisstudio to make this kind of creative production accessible to anyone.
Have a product that needs a video?

DM me or submit your requirement → https://t.co/CgA8EwArBO

Oasis waitlist is open.