Workflow automation for small offices
I hate copying data from one system to another, and in a small business it’s often the owner’s job, so a vacation turns into catch-up work. I set it up so it moves on its own and you get to breathe.
If someone in your office types each job into your scheduling app and then again into QuickBooks, I set it up so the job shows up there on its own, and your office person just checks it.
At an insurance agency it’s the same with the rater and the agency system.
It’s free and takes about 20 minutes, and if automating it wouldn’t pay off, you’ll know that before you’ve spent anything.
The programs may each work fine on their own, and the time goes into getting the same information from one into the other by hand, then cleaning up when the two don’t match.
Then someone fixes the duplicate customers and matches the app’s payments to the bank deposits, all by hand.
Policies that don’t download get typed in by hand, and so do endorsements, certificates and the commission spreadsheet.
Then someone spends tax season emailing clients for the documents they still owe.
If only one person knows how your numbers move between systems, the numbers stop when that person’s out. I turn what they know into steps a system runs, so the numbers keep moving and that person just checks them.
You pick the reporting period and the question it needs to answer.
It pulls the data, applies your rules, flags anything odd and drafts the report.
It’s the one step that still needs you.
Then you check the calendar and chase whoever hasn’t sent their numbers yet.
You move the numbers over, check the formulas and build the report again from scratch.
And if something breaks along the way, you’re the one who has to notice.
Illustrative workflow · the principle behind my work
Every job starts with the free 20-minute check, which is also where we set the exact price if you decide to go ahead. Every offer in full
Bring the one job someone in your office keeps typing into two systems, and I’ll tell you whether it can run on its own, what it would cost, and whether it’s worth it.
Book a free 20-minute checkI set up one of those jobs so nobody has to type it twice, at one fixed price for the work. It’s working in two weeks, you keep the software you already pay for, and you own it.
If that’s more than you want to take on right now, there’s a $500 first step that’s credited toward the full job.
Fix one workflow in fullAfter a client call, the follow-up is already drafted from your notes and waiting for you to approve or fix, so it doesn’t depend on you remembering it at 9 pm.
Second Brain Install in fullIf you built an app with Lovable, Bolt, Replit, Cursor or v0, the Launch Check goes through the five things that tend to go wrong once people start using it, and the Rescue Sprint fixes whatever it finds.
See the Vibe Code AuditThere’s no plan or spec to write first, so just pick the job that takes up the most of your team’s week.
A line or two is enough, like which two systems it goes between, who types it and how often.
You walk me through it the way your team does it now, and I ask questions as we go.
I’ll tell you what I’d build, how long it would take and what it would cost, or why I’d leave it alone.
After that it’s up to you, and if you decide not to go ahead, the advice is yours to keep.
I’m Evan Brown, and I run GIMS Technologies out of Highlands Ranch, Colorado, on my own. The person you talk to at the free check is the one who builds it and checks it before it goes live.
I work in person around the Denver south metro, in Highlands Ranch, Littleton, Lone Tree, Centennial, Englewood, Parker and Castle Rock, and remotely with anyone, anywhere.
More about meMoltbook went viral in January 2026 with its database open to anyone. Wiz counted 1.5 million API keys, 35,000 email addresses and thousands of private messages anyone could read, all because one setting was off. Wiz’s write-up ↗
That setting is the first thing my data access check looks for. Before you launch, or before a lot more people sign up, the Launch Check goes through five things:
My work makes information flow, traceability, and failure handling explicit. I’m increasingly focused on data engineering environments where throughput, reconciliation, recoverability, and orchestration become central concerns.
Below: a thirty-second showreel, results from the lab work, and the projects, each with a path into the architecture.
I designed validation architecture to process LC-MS, GC-MS, and ICP-MS data. Analyses that could take a full day or longer took roughly an hour, giving early-stage labs faster feedback on whether to retest, revise a method, or move forward with a validation report.
The laboratory work behind the softwareWorking software, real screens, and the decisions underneath. Each project has a quick introduction and a path into the architecture.
When the real workflow outgrows the spreadsheet.
Rigid software leaves people managing the gaps in side files and workarounds. GIMS makes the actual process explicit: structured records, configurable workflows, and a foundation that reporting and automation can reuse.
Why it matters The process can evolve without rebuilding everything around it.
Open-source corePython · FastAPI · SQL · React
A result is only part of the story. The Relay captures instrument files, records who did what, and packages evidence for review—so a lab can reconstruct how the work happened.
Why it matters Less handwritten recordkeeping; a traceable record reviewers can check.
FastAPI · Tauri · PostgreSQL · HMAC integrity
Explore the controls & demoI wanted to run more work than I could personally keep in working memory. A shared coordination layer connects parallel agents, records their work, and brings decisions back to me.
Why it matters Long-running work stays organized and accessible across devices.
Agent coordination · MCP · RAG · tmux · Tailscale
See the architectureMoving data out of a database just to reshape it creates another bottleneck. autoSQL compiles a user’s selections into SQL that runs where the data lives, then compares results against the Python path.
Why it matters Numerical disagreements become visible. The performance question stays measurable.
Python compiler · PostgreSQL · generated SQL
Explore the design & evidenceIf a job like that comes to mind, book the free 20-minute check and we’ll go through it together.
If you’d rather write first, talk through a workflow by email
