Building an AI-First Company — Elly Analytics’ Seva Ustinov on AI for Team Productivity | EP78
🎯 Summary
[{“key_takeaways”=>[“Elly Analytics targets the gap in marketing attribution tools that fail to accurately track true revenue from leads for consumer software and service businesses due to delayed backend conversions.”, “The founder transitioned from a successful digital marketing agency to a venture-backed product company, applying deep industry knowledge to build a specialized AI marketing data platform.”, “The core value proposition is building AI agents that analyze marketing channels, creatives, and campaigns to determine true revenue impact, allowing users to improve performance simply by interacting with the system.”, “The speaker heavily advocates for using advanced AI environments like Cursor not just for coding, but as a universal productivity layer that works directly with local files, eliminating constant context switching.”, “Establishing ‘guide rails’ via a dedicated rules file in the AI environment is crucial to prevent hallucinations and ensure the AI adheres to specific structural and tonal requirements for tasks like competitor analysis or PR strategy.”, “AI adoption across the company is driven by finding specific, high-leverage use cases for different roles, such as automating tedious documentation updates for project managers or writing tests for developers.”, “The workflow emphasizes creating a persistent context (files) that the AI can reference, making interactions far more effective and personalized than traditional, stateless LLM chats.”], “overview”=>”Seva Ustinov, founder of Elly Analytics, discusses the company’s mission to solve complex marketing attribution challenges for subscription and service-based businesses where conversions are delayed post-signup. He details Elly Analytics’ transition to an AI-first company, leveraging AI agents within tools like Cursor to automate analysis, rule creation, and operational tasks across marketing, development, and project management.”, “themes”=>[“AI-First Company Transition”, “Marketing Attribution and Performance Data”, “Advanced AI Workflows (Beyond Coding)”, “Personalization and Rule-Based AI Interaction”, “Operationalizing AI Productivity Across Departments”, “The Value of Persistent Context in AI Tools”]}]
🏢 Companies Mentioned
💬 Key Insights
"And a very important part is like anti-hallucination rules because when you work with Cloud Code, especially with Cloud Code, it tends to like write things that are not based on the ground truth because it's very proactive and it starts creating files..."
"The more responsibility and tasks you have, the more you use Cursor because it becomes like a huge leverage for any kind of repetitive tasks, even intellectual, even like managing like 30 integration projects at the same time with dedicated project managers."
"Now, I think our best developer writes pretty much all the code with AI, with multiple windows using Cursor and Cloud Code, and spends something like $1,000 per month on tokens, but the productivity is extremely high."
"Why would I talk to AI that doesn't have all of this information? Why would I do it manually when I can brainstorm on the fly and restructure things just by talking to Cursor?"
"But consumer software businesses, consumer AI businesses, healthcare companies, home services, they have this thing where the website basically generates leads or sign-ups or trials for them, and actual conversions happen somewhere else in the payment system, in the back-end platform. It's delayed in time, so most of the existing tools don't work with that correctly."
"I just ask a Cursor agent to do that for me. It does its magic, downloading files, doing something here. Yeah, it does it without you needing to know how to do it, right? Just say, 'Do it,' and then it does it."