ShieldStone Labs
Austin, Texas, working nationwide. Production software for operations where errors cost money

The technical team for businesses that never hired one.

Most people come to us about AI. Often the honest answer is that they do not need it. What they need is software that works, systems that talk to each other, and somebody who can tell the difference.

$250K+
Processed monthly through a system we built
$1M+
Annual revenue on a platform we built
4
Production systems, four industries

ShieldStone builds and runs the software that owner-operated businesses depend on. Sometimes that means taking a prototype somebody built themselves and making it production grade. Sometimes it means building something that does not exist yet. Sometimes it means being the technical department a company never got around to hiring. The common thread is that the technology has outgrown whoever has been holding it together, and the cost of that is already showing up somewhere.

01

Case studies

Insurance, waste hauling, investment and AI. Four industries, one situation. Something already existed and could not carry the load it was being asked to carry.
Case Study 01 · Insurance

An owner's AI prototype, turned into the system a brokerage runs on.

Based inPortland, OR
SectorInsurance, regulated
AccessRole separated, agency wide
SecurityPasskeys, TOTP, Argon2id
StatusOngoing, three engagements

The agency owner built a working prototype herself using AI. It did what she wanted on her own machine, and there was no version of it that could run a brokerage. ShieldStone took it into production: the full commission lifecycle, financial output reconciled across four independent surfaces, and the audit and security posture a regulated industry requires. It is live and in daily use across the agency, carrying real transaction volume every month. Every account and every piece of infrastructure sits in the client's name and has since day one.

Case Study 02 · Waste hauling

An employee's internal tool, now a platform other haulers run on.

Based inAustin, TX
SectorWaste hauling, operations
Revenue run through it$1M+ / yr
Dumpsters managed130
DatabasePostgres RLS
StatusOngoing

An employee at a dumpster company built the tool his operation ran on. He understood the business better than anyone and had reached the limit of what he could carry alone. He brought us in to make it secure and usable by his whole team, and what that work opened up was larger than the original ask: a multi-tenant backend with real data isolation, scheduling and dispatch and billing unified in one system, and a photo and GPS workflow that gives dispatch an exact pin for every dumpster in the field. It now operates as its own company, and the first operator on the platform is on track to run over a million dollars of revenue through it this year.

Case Study 03 · Investment
Intertwined Investors logoIntertwined Investors

An automated trading system, rebuilt before capital was exposed to it.

Based inPortland, OR
SectorInvestment, trading
Asset classStocks and ETFs
Strategies in parallelUp to 50
StatusDelivered

The firm had spent roughly six months on a build that was unfinished and structurally fragile, with real capital intended to run through it. ShieldStone reviewed it and recommended a clean rebuild, in writing, before the engagement began. The system now ingests live market data, scores each signal for probability of success, and manages position opens and closes without continuous human involvement, with every trade logged against the exact market conditions it fired in.

Case Study 04 · Artificial intelligence

Shipping AI output with no way to tell whether it was improving.

Based inAustin, TX
SectorAI, product
BuiltEvaluation layer
ResultMeasured accuracy gain
StatusDelivered, in use

The product depended on model output and the team had no measurement layer underneath it. Every prompt change was a guess. ShieldStone built the evaluation layer they did not have and put it inside the development path, so prompt refinement runs on measured results rather than impressions. Output accuracy improved against a baseline the team can now see and keep testing against as the product changes.

02

You might recognise one of these

01.

You built something with AI and it works on your machine, but it cannot run the business.

02.

Your team retypes the same information into three systems every day.

03.

One person built the thing everything depends on, and only he understands it.

04.

You are spending money on AI and nobody can tell you whether it is working.

05.

You have no technical person, so every technology decision lands on you.

Owner-operated businesses, generally 5 to 25 million in revenue, complex enough to leave a paper trail. Insurance, waste hauling, trades, property management, freight and similar. There is a system the business depends on, one or two people understand it, and a technical decision nobody internal can make.

03

Who you work with

Ennis M. Salam

Ennis M. Salam, principal engineer

Every engagement is scoped, reviewed and signed off by the same person. When an engagement needs more hands, they work to that standard and nothing ships without going through it. There is no account manager between you and the work. When you ask why something was built a certain way, you get the answer from whoever made the decision.

Previously a software engineer at Cisco and a team lead contracting for Scale AI, leading twenty-three engineers, and an electrical and computer engineering graduate of the University of Texas at Austin. Most of the work on this page arrived by referral from the client above it.

The advice you get on AI will be the same advice either way. Sometimes the answer is that you do not need it, and saying so is cheaper for everyone than building it and finding out.

Connect on LinkedIn
04

Where to start

The Production Readiness Assessment

$10,000 · Two weeks · Fixed scope

Two weeks of engineering and analysis that ends with a build specification, a fixed price broken into modules, and a written account of everything that could move it. If you build with us, the fee credits back against the build. If you read the specification and take it to your own team or another firm, that is fine. You paid for it.

The Build

Fixed price · Fixed scope · Paid on demonstrated milestones

Larger builds are divided into milestones, each with its own scope and price agreed in advance. Every milestone is demonstrated as working software before it is invoiced. You see the thing run, then you pay for it. Nothing about the total moves without a written change with a cost attached to it.

Fractional CTO

Monthly · Fixed engineering hours

Some businesses do not need a project, they need a technical department. Someone who knows the systems the business runs on, handles what breaks, builds the next thing when it is needed, and gives a straight answer when a vendor or an employee proposes something expensive. Most owners have been meaning to hire a technical lead for two years. This is the version that starts next month.

It works as a starting point, not only as something that follows a build.

05

How an engagement runs

From first conversation to a system in production, and what we commit to along the way.

A conversation

Free, and usually short. What do you have running right now, and what is it costing you that it does not do. If we are not the right firm, you will hear it here rather than three weeks in.

The assessment

Two weeks, fixed at $10,000, paid up front. The full scope is in the section above.

Tell us what you're running on.

What do you have running right now, and what is it costing you that it does not do? That is the whole first conversation. If we are not the right firm we will say so on the call.

Get in touch
ShieldStone Labs Software & Product Engineering · Austin, TX PrivacyTerms© 2026