AI consulting and software for growing businesses

Get unstuck.
Then get ahead.

We find the friction that repeats in how your business works, build practical ways to remove it, and make sure your team can run the result without us. Diagnosis, working software, and handoff, from the people who do the building.

Good fits Repeated document work. Systems that don't talk to each other. An internal tool that keeps not getting built. A team that's curious about AI and hasn't seen a result yet.

Talk to us → See the work ↓
// The work

These were the first problems we built for.

One product we operate, and a shelf of tools that came out of building it. Everything on this page is internal work, labeled that way. These are not client case studies. It's what you can inspect before you write to us.

Product · live and paid · internal example

Exec Foundry

Job search tools for senior professionals. The AI carries the volume. You keep the judgment.

The problem
A senior search runs for months. The fortieth application goes out weaker than the first, and every tool we found was built for entry-level applicants.
What we built
A workspace that generates materials, interview prep, and a sourced company brief for each pursuit from one record, and keeps every pursuit's story and status for the duration.
What the user does
Reviews the prep, shapes the narrative, decides when it's ready. The product stops at the part that makes them worth hiring.
One decision
We turned down auto-apply, the first feature everyone asks for. Sending more, faster, was the wrong problem to solve.
Exec Foundry's Resume Forge: a master resume, then Diagnose, Forge, and Verify steps with findings grouped by signal

Resume Forge. Diagnose finds the gaps, Forge proposes edits, the user decides what ships.

1 / 3
Read the story → execfoundry.com →
Open source · internal

ForgeTrail

Recurring problem
Every new AI coding session started from zero and re-fought bugs the last one had already settled.
What we built
Decisions, phase, and lessons kept in the repository, so the next session starts where the last one stopped. When a later job matches a sibling on the shelf, ForgeTrail offers that tool. Decline is success.
Why it matters to a client
The same habit is how we avoid charging you twice for the same discovery. Decisions get written down where the work lives.
forgetrail.dev →
Open source · internal

Cold-eye

Recurring problem
New AI tools land every week, and most of them make sense to the person who built them, not to the next person who opens them. We wanted a stranger's read on all of them, ours included.
What we built
A review skill that reads a finished artifact and returns a readiness verdict with ranked findings. It changes nothing; you decide what to fix.
Why it matters to a client
Anything we build for your team gets this read before handoff. Findings you apply. The skill does not edit the source.
coldeye.dev →
Open source · internal

xFacts

Recurring problem
Adopting a model, tool, or agent meant a source dive to find out what it touches and what it hides.
What we built
Five compact labels, one per layer of the AI stack, with "undisclosed" as a visible value. A validator checks structure. It does not certify truth.
Why it matters to a client
When we recommend a tool for your business, this is the set of questions we already asked about it.
xfacts.dev →

Dozens of tools on the shelf, each with a clear boundary and a label that says what it touches. See our tools →

// The problem

AI is everywhere.
Where do you start?

You've seen the headlines. You've tried the tools and maybe subscribed to a few. But there's a gap between "playing with AI" and making it work for your business.

We sit with you and your team, find the friction that repeats, build the thing that removes it, and teach you how to run it.

Sound familiar?

The same document gets rebuilt every week. Proposals, intake summaries, reports: someone assembles them by hand from the same sources, every time.

Your systems don't talk to each other. The data lives in three tools, and a person is the integration.

There's an internal tool nobody has time to build. Everyone knows what it would do. It's been on the list for a year.

You're paying for AI tools you barely use. Nobody set them up around your work or showed the team how.

Your team is curious about AI, and curiosity hasn't produced a result yet.

// Where to start

Four ways in.
Each one ends with something you keep.

Pick the one that matches where you are. For each: what we look at, what we need from you, what you hold at the end.

DIAGNOSE

Workflow assessment

What we look at
How work moves through your team, the tools you pay for, where the hours go, and where handoffs break.
What you bring
Two or three people who do the work, an hour each, and a look at the tools they use.
What you get
A written shortlist of the problems worth solving, ranked by payoff and effort, with a recommended first build. Short enough to read in one sitting.
Then
Pick one with us, or take the list and run with it yourself.
BUILD

Focused implementation

What we look at
One bounded problem: the top of the shortlist, or the one you already know about.
What you bring
The people who do that task today, and what a good result looks like to them.
What you get
A working automation or tool, success criteria agreed before we start, and a check against them before we call it done.
Then
A handoff that covers what it does and how to change it. Your team runs it.
TEACH

Team enablement

What we look at
The tasks your team does every week, and the AI tools they already have access to.
What you bring
The team, a half day, and the work they already do this week.
What you get
Hands-on sessions on their own tasks, plus the prompts, checklists, and workflows they build during them. Those stay with you.
Then
They use it the following week. We check in on what stuck.
SHIP

Product or internal tool

What we look at
The tool that keeps not getting built, and what has to be true for it to be worth building.
What you bring
The problem, the people who would use it, and someone who can say no.
What you get
Discovery that settles the what before the how, a rough version you can react to early, then the build and the unglamorous hardening.
Then
You own the code and know how to extend it. Exec Foundry was built this way.
Every engagement
  1. Tell us how you work
  2. We shortlist what pays off
  3. Start small, then build
  4. Your team owns it
// Who you'll work with

You talk to the person
who builds it.

Sam Douglas
Founder, Catalyst Forge

Sam has spent his career building software that didn't exist yet: at Raytheon, Choice Hotels, and Verisign, and as co-founder and founding engineer of Mindcorp. Alongside that, a sixteen-year consultancy serving dozens of small businesses in southern Arizona. In addition to founding Catalyst Forge, he builds and operates Exec Foundry, wrote ForgeTrail and the rest of the shelf, and writes the essays.

On an engagement, Sam does the diagnosis, the build, and the handoff himself. When a project needs more hands, we bring in collaborators, and Sam stays the person you talk to. No sales team, no account managers.

Your business drives the decision

We don't start with what AI can do. We start with what your business needs. If AI is the right answer, we'll tell you. We'll also tell you when it isn't.

You leave ready to run without us

The goal isn't a long-term contract. It's that you and your team understand what was built and how to grow it without needing us on speed dial.

// Writing

Two essays that show how we work.

One is about settling the what before arguing about the how. The other is about a feature we refused, and why saying no was easy.

· 4 min read

What Over How

The kickoff ends early. Everyone nodded at modernize. Three people just heard three different projects, and nobody finds out until month four.

· 5 min read

What We Didn't Build

We shipped Exec Foundry, a search workspace for senior professionals. The obvious feature we turned down says the most about how we build.

All writing →
// Let's talk

Tell us what's repeating.

Tell us what's slowing you down. We'll listen, ask the right questions, and tell you whether AI belongs in the answer.

Talk to us → I have a question

Or write to us directly: hello@catalystforge.com

What to include

The work that keeps repeating, who does it, and what you've already tried. A paragraph is plenty. Don't clean it up; the mess is usually where the answer is.

What happens next

You get a reply from the person who would do the work, not a sales sequence. If it looks like a fit, the next step is a conversation about how you work. If it doesn't, we'll say so and point you somewhere useful.