Wedge

The initial beachhead is after-hours emergency dispatch for residential HVAC companies. This niche faces acute pain because staffing human dispatchers overnight is expensive and highly prone to technician misallocation. Once the agent proves reliability in managing low-volume overnight emergencies, it expands to handle the high-volume daytime schedule and eventually controls daily parts ordering.

Timing

Recent advancements in reasoning models allow AI to instantly parse unstructured customer intake calls and categorize the exact mechanical issue and required technician skill. This acts as the missing structured data feed needed to run combinatorial routing algorithms in real-time.

Why This ICP

Mid-sized residential HVAC and plumbing fleets operating 15 to 50 trucks experience severe routing complexity but cannot afford enterprise logistics software. They also experience immediate, measurable revenue impact when a senior sales technician is successfully routed to a high-value replacement job instead of a low-value maintenance call.

Size Of Prize

There are approximately 50,000 mid-sized HVAC, plumbing, and electrical fleets in the US currently employing at least one full-time dispatcher. At an average loaded labor cost of $60,000 per year per dispatcher, the addressable labor spend is roughly $3 billion annually.

Gap Narrative

Mid-sized trade businesses rely on human dispatchers to manually assign technicians to incoming jobs, leading to inefficient routing and mismatched skills. These fleets need an automated system that matches the highest-converting technician to high-value jobs while minimizing drive time. Current field service platforms offer static drag-and-drop calendars, leaving the actual optimization up to human intuition.

Defensibility

The system builds deep workflow lock-in as it becomes the sole engine dictating the daily operational heartbeat of the fleet. It also develops a proprietary data moat by continuously learning individual technician conversion rates, repair times, and skill proficiencies across specific equipment brands. As the dataset grows, the algorithm's revenue-per-truck optimization becomes nearly impossible for a new entrant or human dispatcher to beat.

Why This Thesis

An autonomous Service-as-Software agent perfectly fits this problem because dispatching is fundamentally a continuous constraint-satisfaction puzzle. By completely taking over the dispatch board rather than just providing a software tool to a human, the agent executes faster and without cognitive fatigue.

Hardest Part

Handling real-time, intra-day schedule mutations like job overruns and emergency cancellations while maintaining route density, without forcing a human dispatcher to manually rebuild the board.

Min Viable Scope

Focus strictly on residential HVAC fleets of 15-50 trucks performing single-day service calls. Deliberately exclude multi-day commercial projects, multi-technician dispatching, and inventory-gated routing.

Cold Start Problem

The system lacks the granular, technician-specific job duration data required to schedule tightly without causing late arrivals. Break this by running a historical ingestion pass on the customer's existing database to build baseline duration profiles.

Time To First Value

2-3 weeks; requires a shadow-mode period where the system predicts dispatch outcomes against human decisions before earning the right to execute live schedule changes.

Data Moat Available

true

Technical Difficulty

High

SAM

~$250M-400M US and UK mid-to-large quantitative hedge funds

SOM

~$10M-30M realistic 3-year capture

TAM

~4,000-5,000 global systematic trading firms × ~$150k-250k/yr on execution routing infrastructure ≈ ~$600M-1.2B

Growth Rate

~10-15%/yr, driven by exchange venue fragmentation and the industry shift toward higher-frequency systematic trading models

Paid Comparable Spend

~$150k-300k/yr on legacy multi-asset Execution Management System (EMS) licenses or dedicated quantitative developer headcount maintaining custom routing logic

Kill Thresholds

• Integration time to first live trade > 30 days
• API latency exceeds 25ms during top-of-book volatility events
• Less than 3 paid pilots converted to $10,000 monthly contracts after 90 days
• System-driven trade failures > 0.1 percent of total order volume

Leading Metrics

• Time from API sandbox generation to first live market execution
• Percentage of total fund trading volume routed through the dispatch engine
• Mean millisecond latency per venue acknowledgment
• Number of custom routing rules configured per active fund

What Proves Right

Systematic funds replace their internal Python routing scripts with the API within the first 14 days of integration. Daily trade volume routed through the system exceeds 10,000 executions per active firm with zero missed venue acknowledgments. Firms willingly sign $10,000 monthly contracts after a 30-day proof of value.

What Proves Wrong

Quantitative developers refuse to trust third-party infrastructure for critical execution paths and revert to internal tools after initial sandbox testing. Latency spikes above 50 milliseconds during peak market hours cause unacceptable slippage and render the dispatch engine useless for systematic strategies. Integration cycles drag beyond 60 days due to custom venue API requirements.