.png?width=552&height=88&name=aptitude8%20-%20standard%20-%20White%20-%20LG%20(1).png "aptitude8 - standard - White - LG (1)")
.png?width=750&height=405&name=Building%20a%20Scalable%20HubSpot%20Strategy%20(2).png)
Webinar
Inside the Five AI Agents That Uncovered $1.1M in Hidden Revenue
Hosted by Aptitude 8’s Ana Zhukova
View Auto-Generated Transcript
[0:00:04] By show of hands, how many of you have children? Quite a few. Quite a few. The average American child spends just seven minutes per day in unstructured outdoor play. Seven minutes per day. How insane is that? But according to the World Health Organization, what they actually need is three hours. That's a pretty big gap. Where did that time go? That time didn't just disappear. It got redirected to screens.
[0:00:34] Average child spends seven and a half hours on screens, tablets, laptops, phones, um, which leads to challenges with their mental health and obesity. Playgrounds are the easy answer here, but someone has to manufacture those playgrounds and someone has to purchase th those playgrounds. And that's where the business model in lies. Our client is the largest commercial playground manufacturer in the world. Aggressive brand aggregation is extremely common in manufacturing subindustry.
[0:01:05] Our client is made up of 11 brands acquired one at a time over the years which means that each brand came with its own systems, own tools, own marketing and sales processes, its own way of selling. Our client almost has almost no internal sales reps. Most of the selling outsourced to agencies over 200 outsourced sales agencies layer in 100,000 leads hitting their database every year. And the reality is that nothing was ever built operationally to connect those brands while the portfolio keeps growing.
[0:01:43] As many as 97% of their leads have no outcome, no note, no life cycle stage update, no deal connected. That means that they can't see what's working. So they cannot double down. They cannot see what's uh broken so they cannot fix it. And they cannot see the whole funnel so they cannot grow past that. The funnel is broken top to bottom. So we build five agents to fix the funnel.
[0:02:09] Please meet the context extractor agent. This agent exists because when lead come in they are often missing essential quote information such as age group uh site budget timeline and nobody knows what's missing under when a rep reads it. That's right. Today a live human person reach each one of them and starts chasing and this uh agent s um fixes that problem. First, a Hopspot smart property reads the lead message and separates what's usable from what's missing using three categories.
[0:02:42] Captured, partial, missing. Um, the three questions that actually block a quote. Then we transformed unstructured incoming data in a structured output that both sales team and AI can use. The agent then drops the follow-up email. It asks only for missing information. It won't ask the same question twice. And it's also repeats the back what uh we already know about the inquiry. Nobody asked a school uh district if they have budget when uh they already mentioned that PTO has been fundraising for a couple years.
[0:03:10] For the next agent, I bring you card rescue. A prospect hesitates uh on shipping and leaves the cart. Every abundant card in in this business is worth chasing. Those are five and six figure cards. But if the solution is one person clicking through the list of cards, you will never chase them all. The cart rescue agent chases every single abandoned cart uh and reaches out with personalized me personalized message to get things back on track.
[0:03:40] A generic you left something in your cart doesn't recover $40,000 uh playground. So it writes from what CRM already knows, the pages that they visited, the type of their company they are, uh what kind of build they already collected. As I previously mentioned, leads go to 200 outsourced sales agencies and up to 97% of those leads never come back with notes. These agencies are certainly paid to convert the leads, but uh because they're not full-time employees, they cannot be asked to operate the same way.
[0:04:14] Introducing our third agent, the RAP accountability agent. all of those leads without any notes, any outcome. Rap accountability agents look through all of them, groups them by the sales rep and uh drafts one email, one email to that sales rep to ask for concrete, clear information. It takes that information and and it puts it where it belongs in your CRM and not in the sales rep's head.
[0:04:43] All those agencies and disconnected systems now have context to come in through. Uh but that's just the first step. Now we need to read the reply and update our CRM in a structured way. Those contacts, companies, deals need to be uh associated and connected. So that's where the deal finder agent enters. The dealinder agent reads the reply and extract the outcome, the PO number, the buyer, the value. Then it does something that a workflow cannot do.
[0:05:12] If the evidence is exact, it associates the lead with a deal and writes out come back. If it isn't, it flags for human review so the wrong joint doesn't corrupt attribution silently. The fifth agent, marketing contact multiplication agent. One shared marketing team serves brands, some of which even compete with each other. Demand for content is far past capacity. There is never enough content, let alone personalize the relevant one.
[0:05:45] Each of the 11 brands are set up as brands in a parent uh account. ICP personality writing samples, multiply agent reads the brand kit and creates snippets of content based on persona and brand. A school PTO and a director of a park need different snippets from the same 60 minutes. Clips are just one output. Same source, same brand kit. It will al also create landing pages and case studies.
[0:06:11] The impact here, it's $1.1 million. And that's just in year one. We found it across all five agents. Saved revenue, brought revenue, and that's just the start. Five agent down, more to go. Imagine how much hidden revenue we'll uh find when we build the rest of them. There's no shortage of demand of playgrounds and no shortage of children who would use them. What's in the way is operations and operational problems are what agents are good at.
[0:06:48] Let's give our children outdoor time back. Let's give our children their childhood back. You can go first. This was great and I love the use of the slide. Bonus points there. Uh so you've built five agents. I think you said you had 32 to go. How do you think about QAing the agents and seeing how the quality performs over time and improving them uh based on what you're seeing in the field?
[0:07:14] Uh on each agent right now we have a human oversight. So emails are drafted and verified and we have a system to verify what's uh what needs to be improved. So it's more on iteration right now rather than okay we are letting it run wild and do whatever it need automatically fully uh it is more on a human oversight right now to get to the percentage where we need it to be accurate uh based on the incoming e-message and the output we still iterating on that.
[0:07:42] How did you build trust with your client that these agents were going to deliver these outcomes? That's a great question. there is a huge problem and uh we were looking to solve it. We had proof of concept uh tested it in a safe environment and HubSpot allows you to test without even using uh any credit. So that's honestly a great feature. Uh so after testing it verifying that it's actually possible. Um yeah they were much more comfortable and especially with the layer of human proof I think that's kind of like the first adoption step. You get the proof that this works. the human reviews that once we are comfortable we move forward.
[0:08:22] Tell me a little bit more about the content you're creating with these agents and are you creating proposals or just sort of content to tell about all the playground parks? This is my area of expertise. But uh we in the um content uh agent we um concentrated our attention on the clips.
[0:08:46] So video clips created off the uh larger video uh case studies and blog posts. So these are the top three uh content outputs that we are working right now. And um that's we we have more ideas to go. You could probably solve 32 use cases more, but that's the starting point that we decided to go with. Twice as much time and twice as much budget. What else would you have done with with some of these agents? That is a great question. Honestly, probably iterating, preparing for go for launch, iterating.
[0:09:24] So when we go live and when uh the agents are running we actually delivering you know not 75 but 85% result not 85% but 95% result and that uh why I would do that because that builds the trust for the sales reps for internal uh people who are working on those agents they don't just disregard okay it's another AI tool that doesn't work but preparing a little bit more planning a little bit more testing a little bit more I think that layers on onto the trust and adoption. Thank you so much. Thank you. Thank you.