At Relevance AI, we knew that a one-size-fits-all approach to onboarding simply wasn't good enough. Traditional lifecycle marketing platforms often deliver generic, cookie-cutter onboarding emails that fail to resonate. We knew that we could do better with AI Agents.
We needed an innovative solution that could create a truly personalized experience—one that would engage new users right from the start, without the manual overhead of conventional methods. That's why we built our Lifecycle Marketing Agent, designed to automate every step of the onboarding process and ensure that each new user feels uniquely valued.
The Challenge: Avoiding generic onboarding emails and manual effort
Traditional marketing automation systems churned out impersonal, generic emails that rarely captured the individual nuances of each new user. As a horizontal platform, our new users differ so much that we really don't speak to anyone when we try to speak to everyone.
Basic segmentation never really works. In real life, the industry someone works in makes them much less similar to someone else than, for example, their role. Therefore, the messaging doesn't resonate, and it's really just a big waste of time. When you get down to that level of segmentation, it gets exceptionally complicated very quickly. Existing marketing automation platforms can't do it unless you hack together a complicated spiderweb of IF statements that are impossible to maintain.
The point of segmentation in lifecycle marketing is to be able to deliver personalized messaging but also recommend content for that particular user based on their product maturity and their firmographic profile. The challenge that often presents itself is that new content is always being added—whether it's templates to get started or new guides. Keeping your lifecycle marketing in sync with your content is an impossible task.
The Solution: Building our lifecycle marketing agent
We set out to transform our onboarding experience by creating an agent that automates, firstly, the onboarding communications for a new user. Here's how our multi-agent system works:
Seamless enrichment - no enrichment providers necessary
When a new user signs up, our CRM Enrichment Agent immediately validates and enriches the user's profile. From basic details such as company name to more nuanced information like classifying their role based on custom instructions from us.
It does this simply with Google and LinkedIn—no need for any enrichment services. The beauty of this solution is that it's much more accurate than data enrichment providers, very cheap, and it can even enable us to enrich people who provide their personal email addresses.
Deep contextual research
Next, our dedicated User Researcher sub-agent conducts comprehensive research, gathering insights from company and user LinkedIn profiles, industry news, and specific insights that we are interested in. For us, we really care about any AI initiatives that they or their company may be working on. It then compiles these findings into a detailed research report that highlights key opportunities used later for personalized engagement.
Content recommendation: tool invention and creation
Next, the Agent decides what content to recommend to the user. As an AI Agent company, we can do something that's pretty spectacular. We can generate an AI tool that's uniquely built for them. Leveraging the research report, our generate tool idea tool crafts a tailored tool idea that directly addresses the user's specific job-to-be-done that the Agent determines would be most valuable to the user. The top idea is then passed to our invent tool, which automatically creates an AI tool for that idea.
Setting objectives, giving it personality, and sending the message
All the work done up to this point is all about researching the user and finding—in our case, generating—content that we think is relevant to them. Now we need to design the message and send it.
Each email in our sequence has an objective. In our case, it starts with building your first agent. Each subsequent email has a different objective depending on the stage the person is in in the customer lifecycle. The next integral part is to give the Agent a strong personality. This helps it stand out in someone's inbox and increase engagement. Our Agent has a very strong personality. It's self-referential and even makes jokes.
Lastly, we send the email using an email service that is trusted to ensure high deliverability.
What's next?
We are working on the next iteration of the 'invent' tool, which will actually create AI agents for each specific user. This would enable us to recommend an AI Agent, not just a tool.
Secondly, we'd like for it to cover more of the customer lifecycle. Right now, it's very focused on the first month of a user's experience with Relevance. We'd like to expand this out to cover more of the user lifecycle.
Thirdly, we intend to add more channels to the Agent. The Agent produces a message—there is no reason why that message can't be delivered in-app, as a text message, or as a note that is printed on a t-shirt.
The impact
Implementing our Lifecycle Marketing Agent has completely changed the way we onboard new users. Our process now delivers a fully personalized, data-driven experience that activates users swiftly and effectively. The positive feedback is overwhelming—with many users calling the onboarding emails the best they've ever received.
Watch the lifecycle agent in action
