From Defense to Offense: How BlueGems Automated 65% of Guest Messages and Freed Its Team to Drive Revenue

TL;DR:

• BlueGems wants to scale without increasing its payroll.
• Conduit AI platform was implemented in less than 10 hours by centralizing years of scattered knowledge, thanks to integrations.
• Now automates 60-65 percent of guest messages, avoiding four to six new hires while maintaining service quality.Freed agents shifted from reactive firefighting to proactive revenue generation: upsells, special packages, premium services.
• Three-layer accountability model ensures everyone, rom leadership to individual agents, continuously improves the system.
• AI outperforms humans in 80-90 percent of cases by accessing unlimited historical data; workflows route emotionally sensitive situations to human agents.

BlueGems faced a problem familiar to any growing property management company: the more units it added, the more headcount it needed to handle guest communications. With 185 units and plans to add 12-18 more by year's end, the math was brutal. The company would need to double its six-person team to keep pace, an additional $6,000 to $8,000 monthly in payroll. Worse, every time an agent left, years of accumulated knowledge walked out the door.

The company explored AI tools in mid-2023. When Conduit emerged as an option, BlueGems made a bet: hold off on new hires and see what the platform could deliver.

The Challenge: Knowledge That Disappears

More than just volume for BlueGems, the core issue was the inability to transfer and leverage institutional knowledge. New hires took weeks to get up to speed, learning by osmosis and scattered documentation. When someone quit, their expertise evaporated. The company was trapped in an expensive, inefficient cycle, constantly training people who would eventually leave, taking their hard-won knowledge with them.

Property management runs on details: which unit has the finicky water heater, which owner prefers text over email, how to handle the specific quirks of 185 different properties. To maximize your business, you need this information to be always available, always up to date.

The Solution: AI That Learns From History

BlueGems implemented Conduit, an AI platform that automates guest communications through conversational messaging. The platform handles routine inquiries, extracts information from maintenance requests, manages simple operational tasks, and always learn from every interaction, all without human intervention.

What made implementation fast was Conduit's integration. BlueGems had spent years building knowledge bases across Notion, Google Docs, and spreadsheets. The integration let the company centralize this scattered information and feed it to the AI. In less than 10 hours the system was setup.

The platform now automates 60-65 percent of all guest messages. It pulls from an always growing knowledge base containing years of historical communication data, information a new hire with two weeks of training couldn't possibly match. When a guest asks about checkout procedures, pool heating options, or local restaurant recommendations, the AI responds instantly with property-specific details.

Since implementation, BlueGems has doubled its automation rate from 35 percent to 65 percent over three months by refining escalation rules and improving workflows.

"We've doubled the automation rate from about 35 to closer to 60, 65 over the last three months from changing and tweaking our escalations, improving our workflows."

Shifting From Reactive to Proactive

The real transformation emerged from how the team leveraged their freed-up time. Before Conduit, agents spent their days putting out fires: answering the same questions repeatedly, triaging maintenance issues, handling mundane check-in logistics. The work was necessary but left zero bandwidth for strategic thinking.

"Our team has shifted from defense to offense in terms of guest experience."

With 60-65 percent of messages handled automatically, agents could finally step back and ask bigger questions. What new services could generate revenue? How could they enhance the guest experience beyond basic problem-solving?

• Early check-ins and late check-outs (premium pricing)
• Pool heating services
• Special packages: private chefs, birthday celebrations, anniversary packages
• Holiday additions like Christmas tree delivery

These weren't complex initiatives requiring massive infrastructure. They just required headspace, the mental capacity to think about the business strategically rather than tactically. Before Conduit, the team lacked that headspace entirely.

Making AI Work: Workflows Built on Real Problems

BlueGems runs 12-15 workflows in Conduit. The company's approach is deliberate: build workflows that solve actual business bottlenecks, not workflows for the sake of having workflows.

The risk, as the founder sees it, is falling into a trap where you're building automation that creates busywork rather than eliminating it. BlueGems focuses on the most difficult problems, the ones creating genuine operational friction.

For this company, that problem is maintenance.

How It Works: Maintenance Triage

When a guest reports a maintenance issue, Conduit's workflow extracts three critical pieces of information:

1. A photo of the problem
2. Permission to enter the unit
3. Confirmation that the guest wants someone dispatched

This seems simple until you're managing 50 work orders in a single day. Without this structured information collection, the maintenance team runs around inefficiently, unable to prioritize or schedule effectively.

Workflows also handle smaller tasks at scale: asking cleaners whether early check-ins are approved, confirming late checkout availability, managing routine inquiries that don't require human judgment.

Measuring What Matters

BlueGems integrates Conduit's dashboard metrics into its company scorecard. Each team members tracks individual KPIs: missed calls, messages sent, calls completed. At the department level, the team monitors the overall AI automation rate.

The dashboard's real value is diagnostic. It shows why and when the AI gets stuck. By identifying that maintenance work orders were the top category of escalations, BlueGems knew exactly where to focus its workflow development. The data also reveals time-based patterns, more issues at night versus morning, allowing better resource allocation.

Improving the system is a company-wide responsibility structured across three layers:

• Leadership focuses on strategy and education, networking with other operators, learning from AI implementation experts, and ensuring the team develops skills that will matter in three to five years.
• Middle managers oversee departmental metrics, set automation targets, and ensure the team hits its goals.
• Individual contributors do the daily work that actually improves the system: updating the knowledge base, tweaking workflows, and correcting responses. This layer is critical. If agents aren't gradually improving the knowledge base every single day, the automation rate will stagnate, no matter how much leadership understands AI strategy.

The tool is only as good as the information it receives. Better information requires better processes for identifying what needs correction and implementing improvements.

Quality Control: When Humans Take Over

The biggest pushback against AI guest communication is fear of poor message quality or robotic responses that damage relationships.

BlueGems believes the opposite is true. The AI outperforms human agents in 80-90 percent of situations because it has unlimited access to historical data. A new hire is limited by two weeks of training and whatever they can remember. The AI pulls from years of property-specific knowledge, guest interaction history, and documented solutions.

"The AI is better, in my opinion, than a human agent. If you think about it conceptually, it makes a lot of sense. Because think about a new hire coming into Blue Gems doing guest communication—we're limited to our brains. Now you take an AI tool that has an unlimited knowledge base to pull from two, three, four, five years of historical data."

The 10-20 percent of cases where AI falls short involve emotional intelligence. When a guest shows signs of discomfort, dissatisfaction, or escalation, workflows automatically turn off automated messaging and route the conversation to a human agent. This preserves the human touch for critical situations where empathy and judgment matter most.

What's Missing: Closing the Loop

The next frontier for BlueGems is task management integration. Currently, Conduit extracts information from guest complaints and creates tasks. But then what? The AI still needs humans to coordinate with third parties, maintenance teams, cleaners, utility companies.

The company envisions integrating Conduit with task management tools like Breezeway, monday dot com, or Asana. This would complete the automation loop: guest reports issue → AI extracts information → system creates work order → maintenance team gets notified → task gets completed → guest receives confirmation.

Maintenance remains the biggest bottleneck in property management. Solving it requires a collective efforts across multiple operational tools.

The Bottom Line: Leverage, Not Replacement

BlueGems didn't cut headcount when it implemented AI. The six-person team remains intact. What changed was the nature of their work.

Agents no longer spend their days answering the same questions repeatedly or manually triaging routine maintenance requests. They spend their time building revenue-generating services, handling complex guest situations that require emotional intelligence, and continuously improving the AI system that handles everything else.

The company avoided hiring four to six additional agents, saving $6,000 to $8,000 monthly, while simultaneously increasing its capacity to deliver better guest experiences and generate new revenue streams.

That's the actual promise of AI in operations: not replacing people, but giving them leverage to focus on problems worth solving.

‍