Sriya Maram - Co-Founder of Persana AI: The Modern GTM Stack, Prompt Engineering, and How AI is Impacting GTM

Sriya Maram 0:00

I think we’re going to start entering a totally interesting but new era of digital workers within the workplace—not replacing the workforce, but working alongside it. We’re going to see the nature of work change as well. For me, I’ve always been someone who embraces technology because it’s going to come whether you like it or not. So you can either stay ahead of the curve, educate yourself, and use it to your advantage. That’s why I really want to encourage all GTM professionals to start learning prompting, start becoming GTM experts with AI, and learn how to use workflows and things like that.

Callan Harrington 0:43

You’re listening to That Worked, a show that breaks down the careers of top founders and executives and pulls out those key items that led to their success. I’m your host, Callan Harrington, founder of Flashgrowth, and I couldn’t be more excited that you’re here.

Welcome back, everyone, to another episode of That Worked. This week, I’m joined by Sriya Maram. Sriya is the co-founder of Persona AI. Persona is a Y Combinator-backed AI tool designed to enhance sales through hyper-personalization, seamless integration, and unified workflows.

Prior to Persona AI, Sriya was a product manager at LinkedIn. This was a really interesting and enlightening conversation. We talked about the value of participating in hackathons, and I never thought about it from the perspective of the people you can meet and the real experience under pressure that you’re getting, which Sriya walked us through. We also dove into something we hear about quite a bit—how using your own product impacts your product roadmap. Sriya walked us through some real-world examples where they did that specifically, and some of the areas I found just fascinating. We’ve all heard about prompt engineering and why it’s so important. But we dove into that and how you can improve on it. We also took a deep dive into the modern go-to-market stack—what it looks like now, where it’s headed, and how AI is already impacting it. Sriya provided real applications of how this is working today, walking us through different examples. I think this is a great listen for anyone thinking about growing their business in a modern way. The reality is, this is going to catch up with us extremely quickly.

So with that, I’ll jump right into the show.

Sriya Maram 2:49

One of the places I’d love to kick this off is—tell me about how you met your co-founder.

Callan Harrington 2:54

Yeah, so we actually used to work together at LinkedIn. We were on the same team before. I think what really sparked us working together, apart from our original teamwork, was that we participated in company hackathons together. When you’re working in a hackathon environment, you’re able to ship things fast and see things go live very easily. That made us realize, Hey, I really enjoyed working with you on new ideas. Would you like to do some side projects together? Even before ChatGPT came out, we were tinkering with OpenAI’s API. We were in their beta for that, which was super cool. I think that really led to us starting Persona because we were early movers in the AI space. My co-founder has over five years of ML experience, and combining that with our years at LinkedIn in the B2B data space, it was a great fit for us to start Persona.

Sriya Maram 3:44

I’m actually curious—do you still do these hackathons? You had a long history of running hackathons going back to UC Davis, which looks like it was a pretty big hackathon. Is that fair to say?

Callan Harrington 3:54

Yeah, I love the energy at hackathons in general. Back when I was in university, I got involved with our hackathon team, and by my senior year, I was the president running them. What hackathons do is bring people together in teams, work on the latest technology, and ship solutions. It’s a very creative and collaborative atmosphere. It’s also a great opportunity for people to upskill and launch things. A lot of entrepreneurs I know got their start—or at least their feel for entrepreneurship—in hackathons.

Sriya Maram 4:23

To me—and please correct me if I’m wrong—it almost feels like, yes, you get creative ideas out of hackathons, but more than that, especially in a bigger company, you get to see who is thinking creatively and innovating. Is that fair?

Callan Harrington 4:36

100% right. It’s not just about building a product or prototyping an idea. It’s about how you position it, market it, and plan to acquire customers. At the end of the day, that’s what every entrepreneur needs to do—differentiate in the market. So, yeah, hackathons are a great way for people to collaborate. It’s also a mini way to test whether launching a company together is a good idea because it gives you hands-on experience working with that person.

Sriya Maram 5:00

Yeah, and you’re in this time-boxed, high-pressure environment. That’s when you really get hacky. You’re thinking, Okay, if we can’t build an end-to-end solution, how can we get our MVP out? That’s usually what you build in a hackathon—the minimal viable product to start showing value and getting customers before you build everything.

That’s really cool.

So during your time at LinkedIn, and it looks like this was with your current co-founder as well, you created Viral Copy. It looks like it was a pretty big success—you had 2,000+ users and reached number two on Product Hunt. I guess one of the things I’m curious about is—why didn’t you continue pursuing it?

Callan Harrington 5:40

Viral Copy was more of a horizontal product. It was designed for generic marketing teams, but where we saw AI going was toward task-specific applications. For business use cases, AI needs to fit into workflows and be trained on task-specific data—not just use generic GPT outputs. An LLM is only as powerful as the data that works alongside it. And the best use case we saw—especially given our time at LinkedIn—was how powerful B2B data is. That’s really what sales and marketing teams need. It’s not just about prompting GPT for a blog post or email. It’s about integrating different data points and AI models to work together in a seamless workflow.

Sriya Maram 6:20

So what I’m hearing is that when you created Viral Copy, it wasn’t necessarily built for a very specific task. It was more broad.

Callan Harrington 6:27

Yeah, exactly. Anyone could prompt anything for generic use cases. But let’s say two different companies—each has its own tone of voice, its own data, and its own approach to outreach. So what ended up happening was that my co-founder and I had to manually prompt and guide the user. We realized the main problem was that it wasn’t task-specific to what the user actually wanted to do.

Sriya Maram 6:50

When you say task-specific, do you mean being customized for that company?

Callan Harrington 6:55

Yeah—customized for that company and for who they’re targeting. For example, the way you research and reach out to an executive is very different from how you’d reach out to an individual contributor. Even the signals that matter vary. Some companies care about hiring signals. Others focus on job changes. Some look at funding rounds or tech stacks. Each company’s prospecting process and what’s important to them is different.

Sriya Maram 7:20

That makes total sense. Was the challenge that you couldn’t do all of that in one product?

Callan Harrington 7:26

Exactly. That’s why we evolved into Persona.

Callan Harrington 7:26

Exactly. That’s why we evolved into Persona. We decided that instead of creating a generic AI tool, we would build something that could aggregate data for companies in a way that eliminated the need for multiple lead generation tools.

With Persona, companies don’t need five different subscriptions for lead-gen data. On top of that, we started building AI agents and workflows that help businesses automate tasks. So instead of using five different tools for lead enrichment, personalization, and outreach, you can do it all in one place. Persona enables you to enrich outbound leads, personalize messaging, and send email campaigns autonomously.

Sriya Maram 8:00

That makes sense. What was the moment that convinced you Persona was the right direction? Was it just the desire to start something? Or was there a specific moment when you realized, Oh, we’re onto something big here?

Callan Harrington 8:10

I think it was a mix of both. My co-founder and I were among the early people seeing the potential of AI and how it was going to transform every industry and workflow. At some point, I found myself educating my peers, my higher-ups, and other business leaders about AI—explaining how GPT was evolving and how AI models would shape the future. When I started feeling like a subject matter expert, I thought, Who better to build and launch a company in this space than me? Given that we had experience working in enterprise and startup environments, we had insights into how AI could actually integrate into business workflows. We saw this big opportunity—not just in AI as a concept, but in making it actionable and enterprise-friendly. Even now, AI adoption is still growing. Some companies are using it and seeing results, while others don’t know how to leverage it properly. Some companies tried it but didn’t get the results they expected because it wasn’t specific enough for them. That’s where we saw the opportunity—to bridge that gap.

Sriya Maram 9:15

That makes a lot of sense. I think AI is amazing, and I use it all the time, but there are definitely things I wish I could do that I don’t have the technical skillset for. I’ve heard you talk about this, and Sam Altman has mentioned it as well—jobs are going to change. The tasks people do in their roles will change. Getting ahead in AI and prompt engineering will be crucial. Is Persona designed to eliminate the need for prompt engineers, or does it allow them to get more out of the tool?

Callan Harrington 9:42

It’s a bit of both. When we first launched, we were building templates and playbooks to assist customers—teaching them how to prompt, how to use data effectively, and how to create workflows. Now, we’re seeing a shift. Experts are coming to us with new workflows and teaching us things we hadn’t even thought of. People have gotten really, really good at prompt engineering, and it’s been fascinating to watch.

That said, our platform is designed to be easy for complete beginners as well. We want to make sure users can get value right away. As they progress and get more advanced, our platform is there to scale with them. So whether you’re a beginner or an expert, you can build AI-powered workflows that match your needs.

Sriya Maram 10:20

For listeners who aren’t familiar with prompt engineering, can you explain what it is and how someone can get started with it?

Callan Harrington 10:26

Absolutely. Prompt engineering is essentially learning how to structure your inputs to an AI system so that you get the output you want. It’s like coding—if you give the system clear, specific instructions, you’ll get better results. A good way to think about it is to imagine you’re explaining something to a five-year-old. The more specific you are, the better the results. If you’re asking the AI to research something, tell it exactly what to focus on and what to avoid. Give it relevant keywords and constraints. There are a lot of free courses, YouTube tutorials, and certifications that can help people get started. Many companies—including us—are launching training resources to help users improve their prompting skills. My advice is to start with the basics and experiment. Try different prompts, see what works, and refine your approach.

Sriya Maram 11:15

So you started Persona in the sales tech space, which is already a crowded market. How did you get your first customers?

Callan Harrington 11:21

We were lucky to be part of Y Combinator, which gave us an amazing community to start with. YC has built some of the best companies in Silicon Valley—Airbnb, DoorDash, Stripe—so just being in that environment helped us get initial beta testers and feedback. After that, we used social media, customer intros, and referrals. We also have a creator program, where AI experts showcase different workflows and partner with us to promote the product. Referrals have been a huge driver of inbound interest. And, of course, we use our own product—we run a highly successful outbound motion using Persona itself.

Sriya Maram 11:55

I’d love to walk through an example of how you use Persona for outbound. Can you share a specific campaign?

Callan Harrington 12:00

Sure. One of our most successful campaigns is based on hiring signals. When a company in our ideal customer profile (ICP) starts hiring for sales, GTM, or growth roles, we trigger a sequence. Persona does the research and crafts a message based on that company’s growth, hiring activity, and industry. For example, the AI might generate an email that says: “Hey, I saw you’re hiring for sales roles and growing your GTM function. Your company has grown X% over the last year, and to support that growth, here’s how Persona can help you scale outbound more effectively.” What makes this powerful is that it’s not just generic outreach. Our AI personalizes the messaging based on real-time signals—hiring activity, company priorities, funding rounds, job descriptions, and so on.

Sriya Maram 12:40

That’s really interesting. What channels do you use for outbound—email, LinkedIn, voicemail?

Callan Harrington 12:45

We use a combination. We integrate with all major CRMs like HubSpot and Salesforce, as well as email and LinkedIn automation tools. We also partner with video personalization platforms, so we can include personalized videos in outreach campaigns. Personally, I’ve found that an email followed by a LinkedIn request works really well. But different strategies work for different teams.

Sriya Maram 13:05

When inbound leads come in, do you run them through the same process?

Callan Harrington 13:08

Yeah. We have a webhook that pulls in new sign-ups, runs Persona enrichment on them, and personalizes the outreach sequence accordingly.

Sriya Maram 13:14

Are you manually reviewing the emails before sending, or is AI handling it end-to-end?

Callan Harrington 13:18

We let teams decide. Some want human review before sending, while others run fully automated sequences. Right now, we’re at the point where I trust the AI enough to let it run on autopilot. But that was after reviewing and refining messaging. Once we found a motion that worked, we automated it.

Sriya Maram 13:40

Have you seen an increase in conversions since moving from co-pilot mode to autopilot?

Callan Harrington 13:44

Yes, absolutely. Our customers report saving over 20 hours per month, getting 3x more positive responses, and generating significantly more pipeline. The results have been great.

Callan Harrington 14:05

You definitely want to have multiple domains, especially if you’re running high-volume outbound. If you’re sending a lot of emails, you need warmed-up domains to avoid spam filters.

Email reputation is critical, and you don’t want to risk burning your main domain. Setting up multiple domains and managing inbox health properly ensures deliverability stays high.

Sriya Maram 14:20

Would you recommend that companies start with co-pilot mode before moving to autopilot?

Callan Harrington 14:24

Yes, 100%. Each company has different messaging and workflows that work for them. Before automating everything, you want to test, refine, and optimize your prompts. Co-pilot mode allows you to review and tweak messaging before fully automating. Once you have a sequence that performs well, then you can confidently put it on autopilot.

Sriya Maram 14:40

That makes sense. Since you’re using your own product for outbound, how much does that impact your product roadmap?

Callan Harrington 14:44

A lot! Using our own product helps us identify improvements we might not have seen otherwise. It makes us more user-focused because we experience the same pain points as our customers. For example, if something feels clunky or inefficient, we immediately prioritize improving it. Also, since we rely on Persona for growth, we ensure it works at scale. If we don’t use our own product, why would we expect others to?

Sriya Maram 15:05

That’s a great point. Sometimes, when you’re too close to your product, you might overlook obvious issues.

Callan Harrington 15:09

Exactly. And the best social proof is when we tell prospects, “Hey, we booked this call using our own platform.” That always sparks interest.

Sriya Maram 15:15

This might be a selfish question, but do you think AI tools like Persona will lead to more consultants, or will they become so easy to set up that consultants won’t be needed?

Callan Harrington 15:23

That’s a great question. I think we’ll see something similar to what happened with Salesforce consultants. Initially, Salesforce had a steep learning curve, so consultants thrived. Over time, as it became easier to use, the demand for consultants shifted more toward optimization rather than setup. With AI, we’re still in the early stages. Right now, many companies need help implementing AI, but as adoption grows, the focus will shift toward optimizing AI workflows rather than just setting them up.

Sriya Maram 15:50

That makes a lot of sense. Since AI evolves so fast, do you think companies will always need external expertise to stay ahead?

Callan Harrington 15:55

AI is a compounding technology—it improves exponentially. What took years before now happens in months. Every six months, we’re seeing massive advancements. That’s where consultants will continue to play a role, especially in helping companies implement best practices. But over time, AI will become more user-friendly, and companies will rely less on outside help.

Sriya Maram 16:15

For businesses adopting AI, do you recommend learning to set up workflows themselves, or should they outsource implementation to a consultant?

Callan Harrington 16:20

I always recommend owning your own workflows. Even if there’s a learning curve, it gives you more control over your business processes. That said, not every company has the bandwidth to become AI experts. Some prefer hiring consultants to get them up and running faster. That’s why we built Persona to be easy to use while still offering advanced capabilities for power users. We also have an in-house consulting team that helps companies build workflows and optimize performance. But ultimately, we want our users to feel empowered and creative with the platform.

Sriya Maram 16:50

How do you personally stay up to date with the latest AI developments?

Callan Harrington 16:53

A few ways. I’m part of multiple founder and startup communities—Y Combinator, GTM founder groups, AI thought leadership chats. I also live in San Francisco, which has an amazing AI ecosystem. Aside from that, I follow newsletters, industry reports, and LinkedIn thought leaders. The AI space moves fast, so staying plugged into those networks is key.

Sriya Maram 17:15

What’s next for Persona?

Callan Harrington 17:18

We’ve seen AI shift from chat-based models to agentic workflows. The next phase is fully autonomous digital workers—AI that can execute tasks with minimal human involvement. We’ve had great success with our Persona Quantum Agent, and we’re launching fully autonomous AI agents. These have been in beta, and customers love them, so we’re making them widely available soon. I believe we’re entering a new era where AI doesn’t replace people but works alongside them. The nature of work is going to change significantly, and I encourage all GTM professionals to get ahead by learning AI, prompt engineering, and workflow automation.

Sriya Maram 17:50

When you say agentic, do you mean fully autonomous AI agents that operate independently once set up?

Callan Harrington 17:55

Yes. Right now, AI is mostly used for augmenting tasks, but soon, we’ll see AI taking over repetitive, manual work entirely. For example, I envision AI sales reps that can autonomously research prospects, draft personalized outreach, respond to leads, and book meetings. We’re already seeing AI-powered voice agents handling outbound calls and inbound qualification. Within the next few years, we’ll likely have digital agents representing us in the online world, handling tasks we don’t have time for.

Sriya Maram 18:25

That makes sense. I just saw 11Labs raise a big funding round for AI voice agents. It’s happening fast.

Callan Harrington 18:30

Exactly. Companies like 11Labs are pioneering AI-driven conversations, and we’re seeing massive investments in this space. The future is coming fast.

Sriya Maram 18:35

For companies wanting to adopt autonomous AI agents, what’s the best way to start?

Callan Harrington 18:39

Work with companies like Persona and 11Labs. AI agents still require onboarding and training to fit your specific workflows. Right now, it’s too complex for most businesses to build their own AI agents from scratch. However, setting up AI-powered workflows is much easier. I’d recommend businesses start there—build automated workflows, get comfortable with AI-powered research and outreach, and then gradually adopt autonomous agents as they become more refined.

Sriya Maram 19:05

Can you walk us through a real example of an AI-powered workflow in Persona?

Callan Harrington 19:08

Sure. One of our customers uses Persona to qualify every website visitor in real-time. When someone visits their site, Persona enriches their profile, checks if they match the company’s ICP, and researches their competitors. Then, it drafts a personalized email: “Hey, I saw your company focuses on [X industry]. Based on your competitors, here’s how we can help you stand out.” Another great use case is social listening. If someone posts about AI co-pilots on LinkedIn, Persona can detect it, pull relevant insights, and automatically send a LinkedIn connection request or email tailored to that post.

Sriya Maram 19:35

That’s super interesting. So, companies set up workflows, and when autonomous agents launch, they’ll just execute those workflows automatically?

Callan Harrington 19:41

Exactly. Right now, we’re helping companies build structured workflows. Once they’re in place, autonomous agents will be able to run them without human intervention.

Sriya Maram 19:50

Last question—if you could go back and give advice to your younger self, what would it be?

Callan Harrington 19:55

Dream big. Even if you don’t see people like you in a certain space, don’t let that discourage you. If something excites you, go for it. A lot of people hold themselves back because they think they’re not qualified or don’t belong. But my mindset has always been, Why not me? That thinking has helped me grow, take risks, and build something impactful.

Callan Harrington 20:15

I love that. You’ve clearly demonstrated that mindset, and it’s inspiring. I’m really intrigued by your fully autonomous AI plans and excited to dive in.

Callan Harrington 20:22

Thanks! This was a great conversation.

Callan Harrington 36:00

Thanks for listening, everyone! If you enjoyed this episode, check out the show notes for links to connect with Sriya. If you liked this episode, leave a review on Apple Podcasts or Spotify. See you next week!