A 45-minute HOA email now takes five to 10 minutes, and travel could work the same way
Long Lake says 70 to 80 percent of the infrastructure behind its Nexus AI platform is shared across industries, letting lessons from HOAs carry over to travel.

David Senra · Bringing AI to the Real Economy | Alexander TaubmanA neighborhood manager opens an email about snow removal and finds a full reply already drafted. What used to take 45 minutes to an hour now takes five to 10 minutes to check and send.
The example comes from Alexander Taubman, co-founder of Long Lake, a firm that buys established American service businesses and installs artificial intelligence inside them. One of its first footholds was HOA management, the business of running homeowner associations that handle maintenance, rules, and vendors for subdivisions and condo communities.
The tool is Nexus, Long Lake's shared AI platform. Taubman describes it as sitting in the middle. On one side it connects to AI models, and it is model agnostic, meaning it picks different models for different tasks and sometimes uses several models to finish a single job. It also connects to a company's data, workflows, and software systems.
About 70 to 80 percent of that plumbing is reusable across industries, including model routing, security, and the coordination of AI agents that can carry out multistep jobs. The remaining 20 to 30 percent is the hard part, mapping the particular workflows and messy data of each business. To do that, he estimates about 70 percent of his engineers work in the field across roughly 15 to 20 states, sitting with employees to see what breaks and what helps.
Email shows why the split matters. In HOA work, much of the job lives in long threads with homeowners and boards. A snow question is not simple. The manager may need to compare 10 other communities the company manages, five vendors, what they charge, how residents reviewed them, and what schedules are available, then write to a vendor and propose an answer.
Nexus is designed to pull that material together and, if it can reach all of those records, to draft the response before the manager wakes up. In Taubman's description, that draft is 99 percent accurate, leaving the manager to edit it in his or her own voice. Getting Nexus to draft in each manager's own voice is something the team is still working on.
That same email muscle could transfer. In travel, 55 percent of inbound questions to agents from travelers arrive by email, covering new bookings, changes, and cancellations. The industry also has decades of tangled supplier, customer, and distributor systems. Because Nexus already learned to untangle vendors and schedules for neighborhoods, Taubman says it could have a large effect on travel from day one.
Questions people ask
Does the AI send the email by itself?
No, in this model the human stays in charge. The AI prepares a draft using prices, schedules, reviews, and past threads, and the manager or agent edits, approves, and sends it. Taubman says his companies are adding staff, not cutting them, because removing routine work lets people handle more customers.
Why would software built for HOAs help a travel agency?
Because most office work looks alike under the surface. Whether it is HOA vendors or airline and hotel systems, the platform needs the same secure connections to AI models, the same way to assign subtasks, and the same way to read email. Only the last layer, the specific vendors, rules, and data sources, has to be rebuilt.
What makes the draft accurate instead of just fluent?
It depends on having good business data and a person who checks. The system works best when it can see prior emails, vendor contracts, reviews, and calendars, and when a manager corrects mistakes so future drafts improve. Without that access and review, even a fluent draft can miss the right price, date, or policy.