Photo as the interface.
The whole app starts from one action — take a picture of the notice. Everything else — parsing, scheduling, dispatch, proposals — is downstream from that single tap.
Case Study · 2025 · Mobile · AI
The elevator compliance business,run from a phone.
Elev8 Comply is a shipped mobile AI app built for Precision Lift Co., a Southern California elevator compliance company. Snap a photo of an Order to Comply notice — the AI extracts the deadline, drafts the proposal, tracks the 48-hour clock, and dispatches the right technician. Everything the owner used to do by hand, now handled from a phone.
The Brief
Paper notices. Hard deadlines. One missed 48-hour clock costs a job.
California elevator compliance runs on Order to Comply notices, strict deadlines, and a state-mandated 48-hour advance notice requirement before any work begins. Precision Lift Co. was handling all of it manually — email threads, spreadsheets, and phone calls. Every notice was a race against a clock the owner had to track himself. The rebuild had to do one thing: make the whole workflow start with a photo, and let the AI handle everything after.
- Client
- Precision Lift Co.
- Product
- Elev8 Comply
- Role
- Product Design · Front-End · AI Integration
- Year
- 2025
- Platform
- iOS · iPad · Web · Next.js 14 · Supabase
The Approach
Five moves that shaped the build.
Elevator compliance software mostly looks like enterprise CRM configured with a company name. That wasn't the brief. This was designed as a mobile-first product for people who work with their hands, and it had to feel like something you actually want to open.
Compliance-first, not features-first.
California's 48-hour advance notice requirement is a legal obligation, not a nice-to-have. Every screen shows a deadline. Every deadline shows its clock. No one ever has to hunt for what's urgent.
One product, three roles.
The owner, the office staff, and the field technicians all use the same app. The interface adapts — same design system, different surfaced actions per role. No parallel platforms to maintain.
AI that shows its work.
Every AI-parsed notice includes its confidence score and the specific fields it extracted. If the AI can't defend the reading, it flags for human review instead of guessing.
Ship it end-to-end.
Design, front-end, AI integration, deploy. Same person from the initial photo-capture concept through the last production release. No hand-off. No lost context.
The Moment
Snap the notice. Done.
The whole product resolves around three seconds. The technician or office admin points the camera at a paper Order to Comply notice. The AI reads it — address, deadline, violation codes, work scope, 48-hour flag — and turns it into structured data. The owner sees the ranked deadline in his pocket before the person who filed the notice has walked back to their car.
01Snap
Point the camera at the paper Order to Comply. Any angle, any lighting.
02Parse
GPT-4o extracts address, deadline, violation codes, and 48-hour flags with confidence scores.
03Dispatch
Skill-tag matching finds the right technician. One tap sends them on their way.
The result
Every deadline, tracked.
Once the notice is in, the compliance calendar tracks every California deadline automatically. Color-coded urgency. Auto-filed 48-hour advance notices. Nothing slips.
The Product
Every surface, built for the field.
A tour of the app across iPhone and iPad. The owner runs the whole business from his phone. Office staff live on iPad. The design system carries across both.
01Auto Proposals
Notice to proposal in under five minutes.
The AI takes the parsed notice, pulls the matching scope items from Precision Lift's rate book, drafts the letter, formats the pricing table, and puts it in front of the owner ready to send. What used to take 30 to 90 minutes now takes one review pass.
- AI-drafted proposal letter with client-specific tone
- Scope table auto-populated from violation codes
- Rate book integration — no manual pricing lookup
- 48-hour advance notice auto-scheduled on approval
0248-Hour Watch
California's advance notice, on autopilot.
Every state-mandated 48-hour advance notice tracked from the moment the AI extracts it from the notice. Color-coded urgency across the whole book of business. The owner glances at his iPad in the morning and knows the day's priorities before his first coffee.
- Every deadline surfaced with real-time clock
- Auto-filed 48-hour advance notices on job approval
- Urgent / Due Soon / On Track color states
- Compliance calendar as the default admin view
03Smart Dispatch
Right technician, right job, right skill.
Every violation code maps to a required skill. Every technician has a skill-tag profile. The dispatcher opens the job and the app recommends the best-fit tech ranked by proximity, availability, and skill match — with two alternates one tap away.
- Skill-tag matching against violation codes
- Ranked recommendations with fit percentage
- Proximity + availability calculated automatically
- One-tap dispatch with SMS notification
04One Overview
The whole business, one screen.
The owner's dashboard. Active jobs, proposals out, techs on job, revenue MTD — the four numbers he needs to see, updated in real time. Below that: live activity feed and the pipeline stages. No custom reports, no exports — the answer is on the home screen.
- Four KPI cards refreshed live
- Activity feed — every AI parse, proposal, dispatch, approval
- Pipeline visualization by stage
- Owner-first defaults, admin-level detail one tap deeper
05In the Pocket
Everything above — from an iPhone.
The entire product works in one hand. Field technicians see their assigned jobs. Office staff can approve proposals. The owner can dispatch a technician from a parking lot. Full parity with the iPad experience — same design system, mobile-first defaults.
- Full parity between iPhone and iPad flows
- Camera-first intake for field submissions
- Push notifications tied to deadline urgency
- Offline-safe: pending actions sync when back online
Under the Hood
Built end-to-end.
Next.js 14Framework · App Router
TypeScriptTypes across the codebase
Tailwind CSSDesign system implementation
SupabaseAuth, database, storage, RLS
VercelDeploy target, edge runtime
GPT-4oNotice parsing + proposal drafting
In Detail
Selected screens.
iPhone and iPad, across the primary flows. Every screen shipped — this is production, not a mockup deck.
Have a business that could use this level of automation?
Let's build the next one.
Product design, front-end, and AI integration, treated as one connected practice. Same person from the first sketch to the last deploy. If that's the conversation you're trying to have, this is where it starts.