Conversational AI inside of CRMs like HighLevel has moved from novelty to necessity.
My time working on AI projects at Boeing gave me enough insight into the power and pitfalls of generative AI, before anyone else had even heard that term. I knew that AI would be a cornerstone of sales, because I had built a working AI lead qualification tool for our own business before ChatGPT had even been released. When ChatGPT did launch, I knew it was time to go all in on what became CloseBot.
At this time, HighLevel didn’t have their own conversational AI. It was up to apps in the marketplace to bring lead qualification and booking AI to HighLevel. In fact, over the past several years there have been many gaps that apps in the marketplace have helped fill.
However, HighLevel is supposed to be an all-in-one CRM, right? That brings us to an important question:
With HighLevel being an all-in-one CRM, where do apps fit in?
A recent conversation between Varun Vairavan (CTO and co-founder HighLevel) and Bryce DeCora (co-founder CEO/CTO of CloseBot) offers rare insight into how HighLevel thinks about its marketplace, AI strategy, stability, and long-term ecosystem and where apps fit into that future… in this specific case, where advanced conversational AI platforms like CloseBot fit within that future.
This article distills those insights for agency owners, SaaS builders, and operators evaluating CloseBot conversational AI and other general integrations within the app marketplace.
HighLevel Conversational AI and the Marketplace Strategy
HighLevel’s marketplace has existed in spirit since 2019 through APIs and integrations, even before the formal app marketplace existed. What has recently started to change is the intentional expansion of discoverability:
- HighLevel has periodic spotlight emails where they highlight notable apps
- Marketplace apps with actions show up within the workflow builder
- Apps with good reviews and/or install history show up higher within the app store
Varun explained that they want more accounts to be installing apps. When a user installs an app, the retention rate of that account increases. Moreso, if that app they installed is a paid app, the likelihood of that account sticking around longer sky-rockets. This underscores a core philosophy:
HighLevel grows faster when customers realize value through specialized tools found in the app marketplace
Rather than treating marketplace apps as competitors, HighLevel increasingly sees them as force multipliers, especially when those apps solve problems beyond the core 80% use cases.
Leveraging the 80/20 Rule to Enhance HighLevel CRM with CloseBotThe 80/20 Rule: Where HighLevel Builds vs. Where Apps Like CloseBot Thrive
One of the most important clarifications from the conversation was how HighLevel decides what to build internally versus what remains marketplace-driven:
- If 80% of users need a feature to be convinced to sign up – HighLevel has either built it, is building it, or will eventually build it
- If 20% or less of users need a feature – the marketplace should build it
This applies to many areas of technology, since businesses and agencies across various niches need many different things. But it directly ties to conversational AI.
HighLevel must offer baseline conversational AI capabilities to remain competitive. However, deep conversational intelligence, complex workflows, industry-specific reasoning, and AI employees that replace human roles are not simple features needed by 80% of their users. Those features are needed by the 20%… those who need more.
That distinction matters.
The Facebook Group Admin Work Necessary for Healthy Marketplace Growth
CloseBot is currently the most installed sub-account app in the HighLevel marketplace. This means it is an app installed at the sub-account level, not the agency level. Each install happened purposefully. In the beginning, CloseBot was mentioned all the time in the HighLevel Facebook community.
However, as CloseBot became more popular, we began to see comments and posts mentioning CloseBot getting removed. Even posts that directly asked questions about CloseBot saw comments about the CloseBot app being deleted. Many apps that grow past a certain point have experienced the same thing… posts and comments mentioning these apps begin to be heavily moderated if the app gets popular enough.
Varun has said that there is more work going in to the standard operating procedure of the Facebook admins, to allow a certain level of discussion surrounding the apps. Facebook posts and comments should be allowed to stand as long as:
- They do not lead onlookers to believe that existing features in HighLevel do not exist or do not function well
- The posts and comments remain focused on HighLevel and its integrations
In short, if someone posts a question or comment focused on HighLevel, and an app is mentioned that fills a gap or need in a non-promotional way, that comment should be allowed to stand. These comments help educate group members of apps in the marketplace that provide the depth that’s needed for specific cases.
Why Apps in the Marketplace Thrive Off Depth, While HighLevel Thrives off Breadth
A key theme throughout the discussion was depth over breadth. Marketplace apps that survive long-term are not shallow add-ons. Shallow, non-niche-specific add-ons get overtaken by the primary platform.
Those who want a warm place to thrive in the marketplace must:
- Go deeper than native features ever will
- Move faster than platform-level development cycles
- Serve advanced users who demand precision and control
And this makes sense. 80% of users wouldn’t understand all of the advanced features if they were there… in fact, the added complexity may cause them to cancel the core platform itself. This is especially true for conversational AI, where CloseBot has traditionally excelled serving the 20%.
For those who need conversational AI that’s responsive in common scenarios, CloseBot may not be the best option. It’s easier, after all, to use the all-in-one platform you subscribed to with HighLevel. However, for those doing high volume or high complexity tasks, CloseBot is there.
CloseBot aims to solve the 20% problems:
- Unrivaled uptime
- A/B testing capabilities
- Multi-provider AI fallbacks
- Objective-based reasoning
- Multi-turn state management
- Integration across all conversation channels, including email
- Even cross-CRM capabilities
Not all of these are “checkbox” features, they do not show up on a feature comparison chart. They are architectural commitments to the 20%. And when the 20% finds CloseBot, it helps them enough that they don’t churn out of HighLevel either. HighLevel wants and needs that 20% to find CloseBot.
Copying vs. Innovation in Software Ecosystems
Software ecosystems naturally evolve through iteration. HighLevel acknowledged that innovation is incremental and often inspired by what already works. CloseBot’s name, as well as the names of other apps have appeared throughout their feature request board. When this happens, HighLevel developers ask no questions and they build.
It’s important to understand that copying, to some degree, is common in software. You take someone elses good ideas, massage them to fit your own platform, and launch. Especially if your users are asking for these features. It’s innevitable. But where Varun has voiced concern is in HighLevel’s process and communication. As they continue to grow, they want to maintain a good working relationship with the apps in their marketplace.
Early on it was easy… Varun knew all of the developers building apps. He would go to them if he knew they were building things that HighLevel would likely build themselves. As HighLevel grows and ecosystems mature, transparency becomes essential:
- Identifying when a marketplace app already solves a problem
- Measuring adoption before rebuilding functionality
- Communicating early with developers when overlap is inevitable
The goal is not to avoid overlap entirely, but to ensure developers of marketplace apps have runway andclarity, and customers get the best possible outcome. Varun mentioned that they should have communicated with CloseBot about the use of objectives and other direct matches of the CloseBot platform. Although these replicated features do not go into nearly as much depth, they were inspired by CloseBot directly.
Many apps have experienced the same thing, and Varun states thislack of communication from HighLevel will be improved upon from now on.
Stability, Scale, and Why It Matters HighLevel
CloseBot’s user base of the 20% need high uptime. This is something that has been relatively less important for the 80% using HighLevel’s conversational AI. However, HighLevel has begun to see the impact of downtime on their user base and have outlined a significant internal investment in:
- Service-level uptime guarantees
- Tiered infrastructure prioritization (P0, P1 services)
- Multi-environment deployment pipelines
- Automated rollback and testing across hundreds of microservices
The improvements from these efforts will not happen overnight. They will be gradual, but they are necessary.
Where CloseBot Fits in the HighLevel Ecosystem
CloseBot is not positioned simply as “better HighLevel AI.” It is positioned as deeper cross-platform conversational intelligence. The conversation between myself and Varun made something clear:
General apps in the marketplace that stay surface-level will die
Apps that go deeper will continue to thrive
CloseBot’s focus on…
- Highly capable job flows with A/B test personas
- Hands-off tools like Smart FAQ and detailed reporting
- Robust testing tools
- And so much more
places it firmly in the category of tools that extend HighLevel rather than compete with it. As HighLevel continues to evolve its native AI capabilities, platforms like CloseBot remain differentiated by depth, specialization, and speed of execution.
HighLevel does not see apps in its marketplace as competitors, and apps within the marketplace should not see HighLevel as a competitor. We are all here to work together and provide value together to scaling marketing agencies worldwide.