Sep 25, 2026 · 1 hr 4 min · 17 segments
Should your AI agents have job titles? Today, Daniel, Kris Rudeegraap, Co-CEO of Sendoso, and Jessica Vose, Head of Marketing at Kana, debate the best way to build AI Marketing agents: role-based…
Jessica VoseGuest
Kris RudeegraapGuestDaniel MurrayHostJessica, I wanna go a little deeper too on like- I, like, what, what, where does the human in the loop, like, like, which- where is the, like, the marketer in the loop in this whole, um, connective tissue that the AI is doing for you that is replacing the market? But I know the, a marketer has to be, like, overseeing whatever-
Um, and actually, you know, personally, one of the reasons I took the role at Kana was 'cause I was a marketer, [laughs] and I was terrified the AI was gonna eat my job.
And since coming into Kana, I realized that not only is there more work f- than ever for marketing to review for that reason, Daniel, because there, there's so much review that is required, and in the initial term for everybody here, the training and the coaching of these agents is absolutely imperative.
And we're all hearing a lot about AI slop, right? And how companies are not doing enough training to make sure that what they're producing and what their point of view is and what their brand manifestation is, is unique enough, and that's where the human in the loop comes in.
So for Kana, what we talk about is kind of like three phases of deployment of true agentic marketing, and the first layer, the first phase is get the intelligence.
You know, like, what Kana applications will do is take all of your data, produce enormous incredible insight.
Um, then there's kind of the decision phase, where it's like, here are your recommendations on stop, start, continue in every single one of these applications, and the human's reviewing not only every single piece of insight and saying, "Eh, that feels off." And what we do is we provide an entire audit trail.
Everything is cited and sourced, so you can see where it's coming from and why it's making that recommendation.
What's the recommended next best action there?" And then further down the line for everyone will be, okay, I have trust that this agent can act on my behalf.
I now can see, you know, the insight, the recommendations, and now I can literally hit a button in Kana that says Run All, and it will actually activate and execute through your entire marketing or ad tech stack.
Chris, I know we, we didn't talk about this, but, like, about your journey, but we've said a little bit about the tools you're using to, like, put all this together, and which ones did you decide to, like, buy to ba- to make these agents? And which ones have you, have you built anything in-house that is, like, custom built for Sendoso? Um, like, I, I wanna know, like, what, what tools you're using to make, like, the, this come alive.
Some of the main tools we use, uh, Databricks has, uh, been used as kind of the context layer.
We funnel all the data in to then make these agents have access to everything from CRM data, product data, call data, email data, everything they need to be more smart.
Um, and so at the end of the day, these agents are really just code- coded, um, that then again sit on, um, some of the, uh, Databricks observability layer that we then use for traces and tracking.
Jessica, I wanna go a little deeper too on like- I, like, what, what, where does the human in the loop, like, like, which- where is the, like, the marketer in the loop in this whole, um, connective tissue that the AI is doing for you that is replacing the market? But I know the, a marketer has to be, like, overseeing whatever-
Um, and actually, you know, personally, one of the reasons I took the role at Kana was 'cause I was a marketer, [laughs] and I was terrified the AI was gonna eat my job.
And since coming into Kana, I realized that not only is there more work f- than ever for marketing to review for that reason, Daniel, because there, there's so much review that is required, and in the initial term for everybody here, the training and the coaching of these agents is absolutely imperative.
And we're all hearing a lot about AI slop, right? And how companies are not doing enough training to make sure that what they're producing and what their point of view is and what their brand manifestation is, is unique enough, and that's where the human in the loop comes in.
So for Kana, what we talk about is kind of like three phases of deployment of true agentic marketing, and the first layer, the first phase is get the intelligence.
You know, like, what Kana applications will do is take all of your data, produce enormous incredible insight.
Um, then there's kind of the decision phase, where it's like, here are your recommendations on stop, start, continue in every single one of these applications, and the human's reviewing not only every single piece of insight and saying, "Eh, that feels off." And what we do is we provide an entire audit trail.
Everything is cited and sourced, so you can see where it's coming from and why it's making that recommendation.
What's the recommended next best action there?" And then further down the line for everyone will be, okay, I have trust that this agent can act on my behalf.
I now can see, you know, the insight, the recommendations, and now I can literally hit a button in Kana that says Run All, and it will actually activate and execute through your entire marketing or ad tech stack.
Chris, I know we, we didn't talk about this, but, like, about your journey, but we've said a little bit about the tools you're using to, like, put all this together, and which ones did you decide to, like, buy to ba- to make these agents? And which ones have you, have you built anything in-house that is, like, custom built for Sendoso? Um, like, I, I wanna know, like, what, what tools you're using to make, like, the, this come alive.
Some of the main tools we use, uh, Databricks has, uh, been used as kind of the context layer.
We funnel all the data in to then make these agents have access to everything from CRM data, product data, call data, email data, everything they need to be more smart.
Um, and so at the end of the day, these agents are really just code- coded, um, that then again sit on, um, some of the, uh, Databricks observability layer that we then use for traces and tracking.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.