AdSmash is an AI video ad platform from Smash Labs Private Limited in Bengaluru, India, led by co-founders Vibhu Bahuguna (CEO) and Arush Sharma (CTO). Ads are moving from static images to video, and AI video models can now produce video ads at the volume ad platforms reward. AdSmash is built to make that work for brands.
The team behind AdSmash
Between them, our founders have built and sold a company, run sales and account teams, and led engineering on products at scale. Both spent years at WorkIndia before starting AdSmash.
Vibhu Bahuguna
Co-founder & CEO · LinkedIn
Vibhu runs the company: customers, growth and the creative work the platform produces for brands. Before AdSmash, Vibhu co-founded Rocket Skills, an agriculture-focused edtech company, led its product and distribution to more than 20,000 paying users, and saw it acquired by Krishi Network. After that came two and a half years as Entrepreneur in Residence at WorkIndia, building a sales-pitch intelligence system on language and speech-to-text models and running 0-to-1 projects with the CEO and COO. Earlier roles include Business Head at LimeTray, growing its account management book to a $1 million portfolio, and program manager on OYO's global IoT team in Shanghai.
Arush Sharma
Co-founder & CTO · LinkedIn
Arush runs product and engineering: the agent, the canvas and the pipeline that connects AdSmash to the video and image models. Before AdSmash, Arush led engineering at Multipli.fi, a decentralised finance company, heading a team of 15 engineers that built the Multipli platform. Before that came Apexclub, a learning platform for early-stage founders that Arush co-founded and built from the course platform to the AWS infrastructure, and two and a half years at WorkIndia, rising to lead engineer on the employer product, the company's main source of revenue. Arush's first work on video was a 2018 research internship at C-DAC, training a deep neural network to pick moving objects out of video.
Why we started AdSmash
We started AdSmash because ads are moving from static images to video, and until recently video was the one format most brands could not make fast enough.
Feeds have moved to video
People spend a large and growing share of their time on social apps watching video. On its Q3 2025 earnings call, Meta said video time spent on Instagram was up more than 30% from a year earlier, that Reels had reached an annual revenue run rate of over $50 billion, and that growth in ad impressions came particularly from video surfaces.
Ad budgets follow attention. A static image still works in places, but the inventory that is growing is video: Reels, Stories, Shorts and TikTok. A brand with only static creative is bidding for a shrinking part of the feed.
The platforms want more creative, not less
Meta's ad system is built to choose from far more ads than before. In the post introducing Andromeda, its ad retrieval engine, Meta's engineers wrote that the number of ad creatives in its systems is expected to grow significantly as advertisers adopt generative AI tools, and that more than a million advertisers had used those tools to create more than 15 million ads in a month.
Our view is that this changes what wins. The platform increasingly does the targeting; the advertiser's lever is the creative. The brands that do well give the system many distinct ads to test, keep the ones that work and replace the ones that tire.
Video is the hard format to make
A static ad can be made in an afternoon from a template. A video ad has traditionally needed a script, a shoot, actors, a location, an editor and sound, which means weeks and a real budget for each one. So most brands make a few videos a month and run them until they wear out, while the leaders in their category test many times as many.
That gap is a production problem, not a media-buying problem. It is the problem we set out to fix.
AI video makes video affordable at volume
Generative video models can now render realistic people, products and places from a script and a few images. A shot that used to need a crew can be generated in minutes. That changes the cost of each ad, and with it the number of ads a brand can afford to test: different hooks, actors, settings, languages and aspect ratios from the same idea.
But the models alone are not enough. Used one prompt at a time, they drift: a face changes between shots, a product's shape shifts, and much of what gets generated is unusable. The real cost of AI video is not the price per second of output; it is the price per second you can actually use. Every failed generation costs money and time.
What we built
AdSmash is the system around the models, built so more of what you generate is usable and every ad teaches the next one.
- Brand context first. Your brand assets, products, past ads and competitors are saved once and used in every script, so ads start from your brand instead of a blank prompt.
- Plan before you render. Scripts and storyboards are directed in one canvas, so you approve the shots before paying to render them. The cost is shown before every render.
- Frontier models in one place. Models such as Seedance, Gemini Omni, Grok Imagine and Happy Horse run in one pipeline; the agent picks the model for each shot, and you can override it.
- Human taste where it matters. On the managed service, people review every ad before it ships.
- Performance feeds back. What runs and what wins is tracked, and it shapes the next round of ads.
See AI video ads, AI UGC ads and catalogue video for how each works, or read customer stories.
Work with us
We are a small team in HSR, Bengaluru. See careers for open roles, or contact us to talk about your creative workflow.
Vibhu Bahuguna (CEO) and Arush Sharma (CTO) founded AdSmash in Bengaluru in 2026.