Episode 100: Building the Future of Reverse ETL Tejas Manohar Hightouch

From my perspective, what I see from having worked with thousands of marketing teams at large scale companies is that they're all super busy and overwhelmed, and the same is true for a lot of the partners, consultants, and agencies they work with. I think there's also a lot of inefficiencies in the program due to how busy and siloed everyone is. There's a person operating the Google channel or the Facebook channel that's super busy and can't look for cross channel opportunities.

Hi, my name is Demetri Panici and I'm a content creator, agency owner, and AI enthusiast. You're listening to the AI Agents podcast brought to you by Jotform and featuring our CEO and founder Aytekin Tank. This is the show where artificial intelligence meets innovation, productivity, and the tools shaping the future of work. Enjoy the show.

Hello and welcome back to another episode of the AI Agents podcast. In this episode, we have Tjas Manohar, the co-CEO of Hightouch, a really cool company building the AI platform for marketing teams. How you doing, TJ?

Doing well. Thanks for having me on the show, Demetri. It's really good to meet you and chat about marketing and AI agents coming together.

So to get everyone started, how did you get into AI? Everyone's talking AI nowadays and AI agents. How did you get into the industry and tell us a little more about what Hightouch does?

Happy to. I actually entered AI through the data industry. My background is in building highly scalable data infrastructure and distributed systems. I used to work at Segment, which was acquired by Twilio in 2020. I was a backend engineer there before starting Hightouch. I don't come from a research background, but one of my co-founders studied machine learning in college and I've absorbed some information from him and others we've hired.

I really got fascinated by AI when ChatGPT came out, like a lot of people, and realized this is really cool and will change a lot of things, especially how marketing teams operate.

How have you specifically taken the broad world of marketing and gotten AI agents to work in it? What kinds of marketing does Hightouch excel at the most?

To give context, Hightouch is a relatively young company started over five years ago. We're a fast growing startup with about 300 people based in San Francisco. We serve companies like PetSmart, the NBA, large airlines, hotel chains, and Domino's Pizza. We started with the idea that marketing teams struggle to use data to personalize campaigns because it's cumbersome and many marketers lack technical skills like SQL or Python.

Companies have tons of data in the cloud in Snowflake, Databricks, AWS, Google Cloud, but that data is disconnected from how marketing teams operate. So we built Hightouch, a platform sitting on top of the data warehouse to make data accessible to marketing teams.

When ChatGPT came out, we realized generative AI would change everything for marketing teams. So we created Hightouch agents, an AI platform specifically for marketing teams. When I say marketing, I mean digital marketing like ads, emails, and banners for large B2C companies.

I have a background in digital marketing and was a paid media manager before, so I appreciate that. What would be your sales pitch for marketing agencies or individual marketing managers about how this will change their lives?

Our AI product, Hightouch agents, combines state-of-the-art AI models like GPT, Anthropic, and Gemini into a platform purpose-built for marketers. It understands marketing concepts like attribution and how to grade ads based on efficiency and scale.

We bring unique context to these AI models that ChatGPT alone wouldn't have. For example, if you ask ChatGPT for offer ideas for Domino's, it has no context of your customers or purchase history. Since Hightouch connects to your data warehouse, we bring customer data as context to our AI platform.

You can ask questions like who is likely to respond to a sitewide sale email. We also bring in all your marketing campaigns across channels, pulling creative, metrics, images, videos, and ad spend data. So when AI suggests ideas, it knows what you've tried and how it performed.

We also incorporate private knowledge like brand guidelines, strategy documents, and campaign goals. This makes the experience much better than just using ChatGPT for marketing tasks. Marketers can get feedback on creatives, campaign ideas, and more with full context.

We're building agentic workflows and little agents that help marketing teams with specific tasks like ad reporting, detecting creative fatigue, lifecycle marketing, and generating HTML-ready emails. We plan to build agents for every marketing discipline that are customizable for each company.

Customization is key because enterprise marketing has many nuanced tasks like legal approvals, brand guidelines, and experiment design that vary by company. We build base agents that can be tweaked for specific needs, like providing reports to brands advertising on your website.

Reporting used to be a huge time suck with dashboards full of confusing data. Marketers want insights and suggestions, not just data. AI can interpret data and provide actionable insights, saving marketers hours of manual work pulling data and creating reports.

Today, marketers spend hours pulling data from different systems and assembling reports. Only a small percentage of the time is there real insight. With Hightouch agents, companies create custom agents that interpret ad metrics, distinguish noise from signal, and provide actionable recommendations.

In meetings, instead of staring at data tables, teams get clear insights on what to focus on and next steps. This saves time and improves decision making.

I remember Mondays were reporting days and it felt like a waste of time. What would you have done differently?

I would have templated comparisons like week-over-week or month-over-month and written out observations using fact, understanding, and action. Much of it was pattern recognition and conjecture, especially during busy seasons when clients might not read reports unless very insightful.

Do you have the ability for marketing managers to quickly answer client questions using LLMs accessing data?

Yes, that's the whole platform. Imagine ChatGPT that knows how to interpret marketing data, analytics, attribution, and customer segmentation with full context from campaign data, purchase data, brand info, and goals. It automates data pulling and provides first-pass interpretation and suggestions.

How do you spread the word about this and articulate the value to clients who might be set in their ways or wary of vendors?

Many companies are scarred from marketing tech tools that overpromise and underdeliver. We've seen churn in marketing analytics and personalization tools. At Hightouch, we have exceptional retention with over 90% of customers renewing annually. The key is making it easy to try and proving value quickly.

My main concern would be how to make creative consistently iterative across copy and media assets. What kinds of media assets do you make?

Marketing involves many steps from research to campaign launch. We excel in data and strategy areas, helping plan content, analyze customer data, suggest campaigns, and evaluate ad performance across digital channels like ads, emails, push notifications, and websites. We're building a content creation studio with AI starting with simpler assets like emails, leveraging existing content repositories.

How does your AI help with audience segmentation and targeting?

Marketers face a lot of context switching. We connect to all customer data and can suggest audiences for campaigns by analyzing data warehouse info, generating SQL queries, running predictive models, and testing hypotheses. The AI can interactively answer follow-up questions about campaign reach, timing, and engagement.

What are some favorite industries you work with and any cool customer stories?

We work with top B2C direct-to-consumer industries including retail, financial services, media companies, gas stations, hotels, and airlines. Any company with direct customer contact using digital marketing is a fit.

One financial services client was bottlenecked on data resources for lifecycle marketing. Now marketers use our tool daily to strategize campaigns and experiments. The advertising team is reducing reliance on agencies by moving media management in-house, using AI to analyze ad accounts across channels and find cross-channel opportunities, improving efficiency and uncovering insights.

It's impactful how you're saving marketers time and money. How do you think AI products like yours will impact jobs in marketing?

Marketing teams at large companies are super busy and overwhelmed, as are their partners and agencies. There are inefficiencies due to siloed work and missed cross-channel opportunities. AI will remove monotonous tasks like first-pass reporting, which is easy to prove and works well.

I don't think there will be huge labor market changes soon. Teams will output more and spend less time on tedious work. Creative is still the hardest for AI to replicate, so creative teams won't be replaced anytime soon.

Large companies want to drive more revenue and improve efficiency. Marketing spend is mostly on program spend, so improving spend efficiency and impact is the biggest opportunity.

What's your favorite personal AI tool?

A hidden hack I've learned recently is using Cursor for general AI tasks. It has strong agentic planning, breaking down big tasks, managing to-do lists, and checking progress. It's great for complex tasks like deep research or generating large software pieces.

This is better than chaining prompts with tools like Zapier. Interactive software like Cursor or Cloud Code with agentic planning can handle complex use cases effectively. We've experimented with Claude Code as the backend for Hightouch agents.

Thank you for listening. If you want to help your marketing needs with amazing AI tools, go to hightouch.com. Thanks to TJ for being on the show. Please leave a like, comment, and review on Apple and Spotify. We'll see you next time.