Software Portfolio

Custom software with real intelligence built in.

Web and mobile apps with machine learning, AI and live data integrations — designed, built and shipped by Stoke Ventures.

Machine Learning
AI Features
Live APIs
GPS & Maps
Cross-platform
Real-time Sync
Secure Databases
Payment Processors
Professional UI Design

“Stoke Ventures consists of a knowledgeable team of professionals who take pride in their craftsmanship.”

— Client review

Andrew Lees, President and Founder of Stoke Ventures — Andrew Lees · President and Founder

About

Who we are

We’re a team of software engineers, designers, and product strategists who turn ideas into working software. Stoke Ventures was founded by an entrepreneur and engineer, and that mix of technical depth and startup know-how still drives how we build: we focus on what will get your product to market and keep it growing there. Our developers have experience building complex software, from multi-platform apps to data-heavy systems and integrations, so we can take on hard problems that simpler shops turn down.

Since our founding in 2014, we’ve developed software and mobile applications for startups, businesses, and institutions, including work for UNC and Virginia Tech. We’ve also learned how to build efficiently. We use AI throughout our process to move faster and keep costs down, and every line of code is still reviewed, tested, and owned by an experienced engineer. You get the speed of modern tools along with the integrity and accountability of a team that has shipped real products for over a decade.

Selected Work

Recent builds

Crewline

B2B · Web + Mobile

Field service operations, off paper.

View live app Office demo Crew app Customer view

Open live app

What it is

One system for dispatch, jobs, customers and getting paid, built for HVAC, plumbing, electrical and construction crews. Crewline connects three experiences — an office dispatch console, a crew app built for the field, and a live tracking link for customers — and uses machine learning to run the busywork behind the scenes.

Built with

Machine Learning · AI Natural-Language Search · Live APIs · Live GPS & Maps · Offline-first Sync · In-field Payments

Key features

ML job assignment
For every open job, a model scores each available tech on drive time from live GPS, skills and certifications, weather, workload, job history and priority, then recommends the best fit.
Explainable, self-improving
Every suggestion shows why it was chosen; when a dispatcher overrides it, that decision feeds back into the model so it learns the shop's preferences.
Weather-aware scheduling
Pushes outdoor and rooftop work around storms and heat, and prioritizes AC calls on hot days.
One-click auto-assign
Fills open slots on the dispatch board in seconds instead of a morning on the phone.
AI smart filtering
Ask in plain English (“overdue invoices over $500,” “customers due for a tune-up”) and AI searches years of jobs, invoices and notes.
Field-ready crew app
High-contrast, big-target mobile app with optimized daily routes, checklists, photos, signatures and on-the-spot card payment; works offline and syncs when signal returns.
Customer tracking link
Live map, tech name and photo, real arrival time, plus quote approval and payment with no login or app install.
Live map and reporting
Every truck and job on one map; revenue, first-time fix rate, time on site and quote win rate.

Chase Sunset

B2C · Cross-platform App

Know which skies are worth the trip.

View live app

Open live app

What it is

Chase Sunset scores tonight's sunset and tomorrow's sunrise from 1 to 10 for any location, with the reasoning behind every score. A physics model reads live weather data, and a machine-learning layer learns from users' real ratings, so forecasts get sharper the more people use it.

Built with

Machine Learning · Physics-based Modeling · Live APIs · User Ratings Pipeline · Google Sign-in · Location Search

Key features

Hybrid scoring engine
Each score blends three signals: a physics model of what the weather says, a learned model trained on past ratings, and how comparable evenings actually turned out — with a confidence level.
Explainable forecasts
“What's driving the score” breaks every forecast into factors like high-cloud canvas, low cloud on the horizon, gray cloud overhead, humidity and air clarity.
Show-the-math transparency
A step-by-step “recipe” reveals exactly how each score was calculated.
Rate-the-sky feedback loop
After each sunset or sunrise, users rate what they saw; ratings train the model.
Model insights dashboard
Tracks prediction accuracy, plots predicted vs. actual, shows model version, and supports retraining.
Saved spots and 7-day forecasts
Favorite locations show the next three skies at a glance, plus a sky journal and search for anywhere.