Stephen Brouhard

Operator first. Builder by necessity.

Two decades running real businesses. Now I direct AI agents to build the software they need, and I decide what ships.

See it running How I build

The TextStack homepage: the headline "What a US text really costs." above a calculator set to 1,000 texts a month with Telnyx, showing $8.17 a month including carrier fees, checked 2026-10-15.
TextStack at textstackhq.com

See it run.

Alderline Agent

Missed calls, texts and web forms for my detailing business land in one owner inbox. The assistant handles routine replies, flags what needs me, and I can take over any thread or confirm a booking by hand.

Shown here: the public demo, with sample data. Open the demo

Alderline's owner dashboard, public demo with sample data: a daily digest of needs, bookings, callbacks and quotes, an inbox of customer threads tagged Manually confirmed, Warm lead, Callback requested and AI handling, and an open thread where the owner manually confirms a Saturday booking.

SSHHIP for Android

SSHHIP puts your own machines' terminals on your phone. I'm building the Android app. Here it runs a shell, opens the Command Dial's pane controls and splits the screen into two panes.

A debug build on my phone, connected to a demo account. More on this project

SSHHIP on Android connected to demo-box, showing recent git commits and the files of a website repository in a terminal. The Command Dial open over the terminal, showing its Herdr submenu of pane controls around a back button. The terminal split into two panes, the git log above and a calendar with disk and CPU details below.

TextStack

What a US text message really costs across eight SMS providers. Every price is read off the provider's own page and re-checked daily. When a number can't be read, it says "checking" instead of guessing.

I own it and direct the build. textstackhq.com

TextStack's price table for Telnyx, Twilio, Plivo, Bandwidth, Vonage, SignalWire, Sinch and ClickSend, with outbound SMS, inbound SMS, MMS and carrier fees, each linked to the provider's site and checked 2026-10-19. Two cells read "checking".

PackSift

A CLI that compares the npm tarballs people actually install, not the Git diff. It lists what changed and leaves the judgement to the reader. Here it compares two lodash releases.

I created it and maintain it. PackSift on GitHub

$ npx @ssbrouhard/packsift [email protected] [email protected]
packsift  [email protected] -> 4.17.21

-- Flagged ------------------------------
  No notable supply-chain signals.

-- Files --------------------------------
  added    5
  removed  0
  changed  12

  C  README.md                    (+2 / -2)
  A  _baseTrim.js                 (new, 0 -> 444 B)
  A  _trimmedEndIndex.js          (new, 0 -> 515 B)
  C  core.js                      (+1 / -1)
  C  core.min.js                  (+24 / -25)
  A  flake.lock                   (new, 0 -> 963 B)
  A  flake.nix                    (new, 0 -> 459 B)
  C  lodash.js                    (changed, binary/large, 530 KB -> 531 KB)
  C  lodash.min.js                (+126 / -125)
  C  package.json                 (+1 / -1)
  C  parseInt.js                  (+1 / -1)
  A  release.md                   (new, 0 -> 2 KB)
  C  template.js                  (+21 / -0)
  C  toNumber.js                  (+3 / -5)
  C  trim.js                      (+2 / -4)
  C  trimEnd.js                   (+3 / -5)
  C  trimStart.js                 (+1 / -1)

Governance Context Object

A reference implementation that stops a delegated AI agent from claiming authority its parent task never had. In this demo a child agent asks to widen "read summarize" into "admin delete". Naive delegation allows it; the GCO runtime refuses.

I own the project and direct the build. GCO on GitHub

$ python examples/demo_authority_escape.py
Delegated child requested widened scope: admin delete
Naive ungoverned delegation allowed? True
GovernanceRuntime decision: Decision(allowed=False, reason='tool authority scope expanded', error_code=<DerivationError.TOOL_AUTHORITY_EXPANDED: 'tool authority scope expanded'>, child=None, value=None)

Selected work

Plus 41 merged pull requests to six projects other people created, and smaller tools and sites.

The sites my businesses run on

The Flagship Detailing homepage: "Premium Mobile Detailing for Cars & Boats, Delivered to You" over a dark photograph of a sports car in a driveway, with Get a Free Quote and View Services buttons.
Flagship Detailing, my mobile detailing business. flagshipdetailingco.com
The Alderline Systems homepage: "Recover the jobs hiding in missed calls, texts, and web forms." beside an illustrated workflow panel of a missed call, a web form and a booking.
Alderline Systems, the company behind Alderline Agent. alderlinesystems.com
Stephen Brouhard, smiling, in black glasses and a black sweater, outdoors against blurred green leaves.

How I work

I run Flagship Detailing, a mobile detailing business in Northeast Florida. Most of what I build starts there: a missed call that never became a booking, an SMS assistant that began asking customers for details they'd already given.

AI agents write most of the code. I name the problem, set the bar for done, read the evidence and decide what ships. Merged, deployed and working are three different claims, and I check each one.

The agent tooling is largely other people's work: Kun Chen's Firstmate and Grok Ship, and methods from Lauren Tan's pstack. I contribute back where I can.

Stephen BrouhardJacksonville, Florida

Read how I build

Building something that has to work?

I'd like to hear from founders, engineers and operators working on agentic systems, developer tools and practical automation.