Enterprise-Grade Engineering
Real-time operations systems, data platforms, and AI in production, for businesses that have outgrown off-the-shelf. We choose the stack that fits the problem, not the one we're comfortable with, and we build it to production standards. Enterprise capability, without the enterprise price tag or the six-month timeline.
How We Work
Enterprise-grade engineering. Small-business economics.
TNC Software is a US-based engineering consultancy. We build the systems that run operations, not the ones that report on them afterwards: edge compute and real-time sync, SQL data platforms, AI wired into live workflows, mobile capture for crews in the field, route optimization, and the integration layer that ties all of it to whatever you already run.
Which of those we reach for depends entirely on what you are solving. A live shared tally across devices needs a fundamentally different architecture to a nightly reconciliation, which needs a different one again to a field app that has to keep working with no signal. Choosing correctly is most of the job, and choosing out of habit is how projects end up expensive and disappointing.
We work in fixed-price phases, each ending in a working system you can use. No retainers, no lock-in, and no per-seat licence that grows as you hire. You own what we build.
What We've Built
Every one of these sits on a different foundation, chosen for a reason we can defend. Names and figures are withheld where a client has not cleared them for publication.
Cash-in-transit · Two regions · Live
When a crew returns to base, nobody could say for certain what had come off the truck. Two people were counting the same sealed bags on two devices with no shared view, and a bag left on board was not discovered until the paperwork failed hours later.
Their existing field app syncs on an interval, so a scan reaches another device only after two sync cycles. That makes a live shared tally impossible by design. No amount of tuning fixes it. We built the reconciliation layer on edge compute with one persistent connection per depot, per day, so a scan on any device appears on every other screen in under a second. The field app was left exactly where it was, and remains the system of record.
ATM cash operations · 22 routes · In delivery
Two dashboards on a hosted reporting tool, an external partner requesting route changes by email, and a source system that hard-deletes its own history on a rolling 30-day window.
Reporting had to move off the source system entirely, because there is no backfill past 30 days. Once it is gone, it is gone. We stood up a SQL warehouse fed by a ten-minute pipeline, so history now accumulates independently of whatever the source does with it. Multi-tenant from the first commit, so onboarding a second customer is a configuration row rather than a rebuild. Building it also surfaced a live data fault the incumbent reports had been quietly reproducing for years: records attributed to the wrong customer, in both directions.
Route scheduling · Multi-country · Live
A cloud route optimizer returns an ordered list of stops. It does not return a schedule an operation can actually run: crews with certifications, vehicles with capacity ceilings, sites with hard time windows, and stops that must happen in a fixed order.
The optimizer solves the geometry. Everything above it is custom, and that is where the engineering lives: the constraint layer, a crew and vehicle allocator that respects hard rules, and automatic handling for the days that go wrong. The whole pipeline is idempotent. Re-run it on the same day's data and it produces the same answer, which is what makes it safe to schedule and then leave alone.
Consumer PWA · 24 origin countries · Live
Put a language model in front of a public, unauthenticated web app without handing anybody an unmetered bill.
The API key never reaches the browser. A Worker fronts it with bot verification, a hard daily spend cap, per-IP rate limiting, input validation, response caching and graceful degradation, and routes requests across two models by cost so the cheaper one handles whatever it can. Where AI was the wrong tool (40,000 postal codes of sales-tax data) we bundled the dataset and did the arithmetic in the browser instead. That path is free, instant, and cannot be wrong.
Low-code has a place, and we still use it where it genuinely fits: capture at the glass, in the hands of a crew. It becomes the wrong tool the moment you need live shared state, real analytics, or access for someone outside the business. Knowing where that line sits is what the four systems above have in common.
Sanitized code samples of the patterns above, and the technical writeups behind them.
What We Do
Live shared state across every device on the job. Edge compute and persistent connections, so an action on a truck is on the office screen before the driver looks up. Not a sync interval. Actual real time.
SQL warehouses fed by scheduled pipelines, so your history survives whatever your source system does with it. Dashboards, drill-down, and exports that reconcile to the cent.
Language models wired into real workflows, behind server-side keys, hard spend caps, rate limits and caching. The controls a live AI feature needs, plus the judgement to say when AI is the wrong tool for the job.
Apps for crews, drivers and technicians. Barcode and seal scanning, photos, signatures and GPS, with an offline queue that reconciles itself when the signal comes back.
Two-way APIs, OAuth service accounts, webhooks and idempotent pipelines that connect the tools you already run. They can be re-run without side effects, which is what makes them safe to schedule and forget about.
Constraint-based routing on Google Cloud Fleet Routing: capacity ceilings, hard time windows, stop consolidation, precedence, and crew and vehicle allocation under rules that cannot be broken.
Who It's For
Job management, quoting, scheduling, and invoicing for plumbers, electricians, HVAC, and maintenance companies.
Mobile apps for technicians, drivers, and field reps. Capture data, photos, signatures and GPS, and sync it all back to the office.
Know what you've got, where it is, and when it needs attention. Equipment, stock and vehicles, all in one place.
Scheduling, bookings, availability tracking, and billing for hire companies and logistics operations.
Track leads, manage customer relationships, and automate follow-ups without paying for bloated enterprise CRM systems.
Checklists, scheduled inspections, audit trails, and compliance reporting for regulated industries.
Why TNC
Get In Touch
Start with the problem, not a spec. The first conversation is about what your operation does today, where it breaks, and what it would be worth to fix, before anybody says a word about technology.
From there we scope a first phase with a fixed price and a working system at the end of it. No hard sell, no jargon, and no six-month discovery. If we're not the right fit, we'll say so.