For small businesses · Cilantric Services
I build digital employees.
The ones that run your company.
A working AI assistant for your business, built on Claude Code and wired into your own context. Not a chatbot. A new employee—one that reads your email, knows your projects, drafts your replies, tracks what you owe people, and tells you what's on fire before you ask.
In production for over a year. Not a demo.
What it actually is
A new employee, not a tool.
One that reads your email, knows your projects, drafts your replies, tracks what you owe people, and tells you what's on fire before you ask. It runs on the same substrate I've been running for my own business for over a year, tuned to how your company actually works—not to some generic template.
The rule I build by
I will never install a set of light switches that stop working when your internet goes down.
That's been my rule in home automation for years, and it's how I build your AI. Progressive enhancement: you own the substrate, and vendor services are enhancements you adopt when they get better, never dependencies you're stuck with.
You aren't locked to one vendor. When a better tool shows up in one area, we use it. When a vendor deprecates something, we swap it out and your workflow doesn't notice.
Nobody else's roadmap decides how your staff work. Gmail's AI can't be an enhancement for you, because it is the platform. Anything built on it inherits its outages and its decisions.
Real example: this summer, an AI company killed off a tool I relied on, basically overnight. Because I build these to swap pieces in and out, I fixed it in a few minutes and none of my clients noticed. If everything had been wired to that one company, the same change would've meant a week of painful rework.
Why me
I broke it first.
I've been running this for my own life and business for over a year. Not a demo—production. It reads my mail, runs my books' cadence, tracks a legal file with a court date, watches my backups.
I know what it catches, and more importantly I know how it fails, because I broke it first. People have tried to build this on other tooling and found out the hard way which parts don't hold. That's the part you can't download.
How it works
Start with one pain point.
01
Start with one pain point
One thing that eats your week, automated properly. (A common first build: inbound email triaged into real tasks, with a link back to the source.)
02
Fixed-price build
Scoped up front, delivered, handed over.
03
Then monthly check-ins
Not one-and-done. The AI underneath changes fast, and keeping it working as it shifts takes someone who knows how it's built.
What it costs
You're hiring, not buying software. So it's priced like a hire: monthly, no signing bonus.
A part-time hire
$30–40k/yr
$2,500–3,500 a month, plus payroll, plus onboarding, plus the risk they leave.
A digital employee
$12,500/yr
No payroll, no onboarding drama, no notice period. It doesn't quit.
- $500 to start
- Covers discovery and getting your context in.
- $1,000 / month
- The build, the check-ins, and keeping it current as the tools change.
- 3-month minimum
- Then month to month. Long enough to prove it, short enough to walk.
First two clients only
$500/month for the first three months, in exchange for a named case study and a reference. Then it goes to $1,000. This expires after two.
What I don't do
I won't come figure out your existing Cowork, your ChatGPT setup, or someone else's half-built project.
I have specific tools I'm expert in and strong opinions about how this is done. That's what makes it a product instead of a favour.
If your setup is held together with copy-paste, spreadsheets, and hope, that's the mess I like walking into.
Send me the ugly version and I'll tell you straight whether AI actually solves it, or whether you just need something fixed.