Like many co-founders, AdSkate‘s CEO Akaash Ramakrishnan and CTO Shreyas Venugopalan have followed a winding path to their AI advertising startup’s current strategy.
Founded in 2019 and originally focused on contextual ad targeting, the Pittsburgh-based company shifted its focus to creative analytics and audience analysis for marketers in 2023. This pivot leveraged the founders’ experience in media, platforms and AI.
Mr. Ramakrishnan worked at automotive agency Team Detroit, now known as GTB.
Mr. Venugopalan the company’s CTO, completed his PhD from Carnegie Mellon and spent nearly a decade – including his postdoc – in Iris recognition and robotics.
With profitability nearing break-even and $2 million in pre-seed/seed funding to date —which includes backing from a Carnegie Mellon investment fund—the 13-person company plans to raise another round of funding in the first quarter of next year.
In the meantime, the firm and its 20+ clients are preparing for the Q4 holiday selling season as creative is being supercharged by AI across ad platforms everywhere.
It’s notable that AdSkate is offering a creative analytics and audience solution that operates across ad systems, including the often opaque ‘black box’ of AI-enabled platforms. Nevertheless, the company believes there is a clear opportunity to work across all of these systems.
Yesterday, tipsheet spoke to Mr. Ramakrishnan about his startup including:
- What problem is AdSkate solving?
- How AdSkate’s AI works with ‘walled garden’ AIs
- Use cases
- Thoughts on AI and the need for large volumes of creative
- The ideal AdSkate customer
- Distribution of the company’s products and MCP
- Client advice: Evaluating AI ad startups
The transcript has been lightly edited for clarity.
tipsheet: What problem is AdSkate solving?
Mr. Ramakrishnan: We enable the advertiser to understand their creatives better by breaking the creative down into its elemental composition. So, it’s different aspects such as image, tone, calls-to-action, color, audio style, music and many other components.
AI gets applied and layered on through that whole journey as we implement different models for different aspects. For example, there is a model for object recognition. There’s a model implemented for music, for images and so on.
How do you think about the interaction between your AI and the AI of other ad systems that your marketer clients are using?
I was in a conversation with a client this very same topic yesterday. Our AI system, in a nutshell, ends up being neutral.
Most of the AI systems that are created by, let’s say, the “walled gardens”, are about fundamentally pouring more money into the campaign that you run in that specific platform – so essentially [optimizing for] a higher ad spend.
The other aspect is that they do not provide the granularity of detail that a marketer is looking for. So that individual granularity is where AdSkate comes in. We essentially are agnostic when it comes to the different platforms that exist in the ecosystem, and we’re able to collect a cohort of information that is applicable across all their marketer’s campaigns and optimizing the journey of the campaign.
What walled garden platforms are you connecting to today?
The usual suspects – Google, Meta and also the other social platforms. Through studying these different platforms, what we have understood is that, fundamentally, the “walled garden” platform has to focus on the bottom line of the revenue and also top line, too. So keeping that in mind, there are a few aspects that the “walled gardens” either miss out on providing or or essentially “pitch it for the future
For example, a brand would like to have a lot more vision into their campaign, and that is not provided in certain platforms. By running with us, since we’re agnostic of all the platforms and the ecosystem, we’re essentially providing a depth of granularity that they would not be getting otherwise.
Can you talk through a use case for your product?
The biggest use case is that if you’re running an ad campaign with five or more creatives, we provide insights about your campaign much more than what you would have ever thought about. There’s also a precursor to this, which we recently launched. We call it “audience analysis” where even before you have launched a campaign, you can put the creative into our platform, and we provide you analysis on the creative itself.
We call that inferred audiences where we infer the creative to tell you what specific audiences that it does well with.
So from a hypothetical customer perspective, a brand can put in the creatives that they have into a “draft” in order to get a sense for the audience the creative would work well with. And then, we can do campaign analysis for the brand, where we can run the analysis based on the platform of their choice, and then provide them detail about the campaign and the creative itself.
Regarding “draft,” are you running against real inventory or is there some sort of simulated experience?
From a pre-campaign perspective, what we’re talking about is synthetic audiences.
What we have done is studied different ethnographic and demographic audiences. And for each audience, we have studied it for different aspects and cultural nuance and even demographical nuance too. So when you run the creative – a static image or a video – you’re getting a much more in depth understanding of whether that ad works well.
What’s the most common metric that your clients are optimizing for today?
It depends on the type of campaign that they’re looking at, but usually they look at click through rate. They’ll also look at conversions. People have gone beyond just looking at impressions – in video, a lot of marketers are focused on video plays and also the number of seconds people are spending on their videos.
AI-enabled, automated ad systems appear to encourage tons of automated creative in order to find the conversion the marketer is looking for. Are you seeing that today in terms of the dynamic capabilities of AI?
I would say it’s about building smarter creatives, better. I wouldn’t say that having a mass quantity of creatives is better, but having a set amount of creatives is good and enables better engagement.
I think one more important factor is to humanize the audience. And to do that, it’s important to understand the audience from the onset.
Who are your customers today? Agencies, brands, performance marketers?
It’s performance marketers focused on lower funnel campaigns – and agencies and brands. What we have realized is that people run different types of campaigns, so it works right from brand awareness to remarketing.
(Mr. Ramakrishnan clarified for tipsheet that AdSkate uses a CPM-based model or subscription for clients.)
Any plans for distributing AdSkate’s products?
There’s a huge opportunity ahead for us in having Model Context Protocol (MCP) in place – this will open up the market for us.
Agencies are becoming much savvier than even a year ago and are trying to have a one-stop shop. Meaning, if an agency subscribes to Anthropic’s Claude, they would like all of the platforms that they use to flow through Claude such that the workflow for the agency user is much simpler.
We felt integrating our solution into Claude, Gemini and ChatGPT was an imperative for our company. And, it’s likely going to be imperative for all the ad technologies that are being used today.
For clients that are evaluating AI advertising startup solutions, how would you advise them to approach understanding a startup’s AI capabilities?
I would tell clients to look at the simplicity of the solution. Simplicity is where it’s at.
If a solution can answer a question in a manner which is easy enough for someone to digest, then that solution has a lot of value, power and impact. If you think about it, even OpenAI’s ChatGPT is a simple chat window. You can ask any question under the sun.
Same goes for AI ad startups except they use the advertising schema.
Follow AdSkate (@adskate) and tipsheet (@tipsheetai) on X.