How we
actually work.
We listen first, ship in weeks, not quarters, and hand back something your team can confidently run.
30-minute intro call · No deck · Your data stays put.
a Studio, not an agency. Senior engineers only, working in your stack.
Median experience
Minimum per automation
SLA
Per-token AI bills
§ 01 / WHAT WE BELIEVE
Four principles,
no slogans.
Process, then product.
An initial call to test feasibility, then a hands-on SystmFlow workshop to map every step, touchpoint and minute of the real process. If automation is the wrong answer, we say so before you spend a penny.
The partner who pitches, builds.
Every person on your project has shipped production solutions for a decade or more. The partner who pitched is the partner who builds. Teams of 2–3.
Automation. Inference. Data. Portal.
Every Systm is four independent layers running on hardware inside your network: workflow automation, local AI inference, your data, and the Systm portal that binds them. Each layer scales on its own.
Yours where it matters.
Hardware on your LAN. Data in your databases. The portal is the SaaS layer that ties them together - you subscribe to it like any service. The foundation is yours; we just keep the lights on.
§ 02 / HOW A PROJECT RUNS
Workshop to SLA.
Six phases.
A single automation runs 2 weeks to 3 months end-to-end depending on complexity. Multiple automations run in parallel.
Evaluate
A working meeting to test requirements, feasibility and rollout reality. By the end we know whether the automation pays back, what the guardrails need to be, and roughly how big the build is.
Map
A hands-on workshop using SystmFlow - our in-house mapping tool - to capture every step of the process, the software touchpoints involved, and how long each event actually takes. That data shapes the build.
Prototype
We build the first working automation in n8n, combining custom code, LLM calls, API integrations and database operations. Guardrails (human approval at critical phases, retries, alerting) are wired in from the start.
Demo
We run the prototype against real cases in your environment and walk it through with the people who own the process. Feedback comes back the same day - we want the awkward edge-cases now, not after launch.
Harden & Roll out
Iterate to consistency and reliability - runs the same way a thousand times in a row. Then rollout: documentation, training for the team, and a quiet hand-over into daily use.
Operate
Every shipped automation comes with an SLA and is monitored from our side - runs, error rates and latency. You get a single line to support and we get the dashboards. No surprises.
§ 03 / THE STACK
Four layers.
One Systm.
The Systm sits on top of the tools you already use. Day one, it captures, triggers and augments. As it matures, it can quietly replace what it once supplemented - on your timeline, not ours.
Automation
n8n at the core, running in containers on mini servers under Debian. Custom code, LLM calls, API integrations and database operations orchestrated together. Open source for small builds; clustered when you scale.
- Visual editor
- Self-hosted
- LLM + API + code
- Scales out
Inference
One or more dedicated machines - Mac Studio, Nvidia GX10 or similar - sized to your workload. Every LLM call runs here, on your LAN. No tokens leave the building.
- Local LLMs
- No per-token bills
- Sized to workload
- On your LAN
Data
Postgres, vector store, queues - whatever the build needs - running in containers on the same mini-server fleet. Run history, audit logs and reporting live here. Encrypted at rest, backed up where you tell us.
- Postgres + SQL
- Vector store
- Audit + run history
- Encrypted at rest
Systm Portal
The web app binding the other three layers together. Local chat against your inference, one-click triggers for automations, reporting and runs history. Delivered as SaaS - we host, maintain, upgrade and secure it; you pay a predictable monthly fee for as long as you use it.
- Local chat
- Trigger automations
- Reporting & runs
- Role-based access
§ 04 / THREE WAYS WE WORK
Pick the shape that fits.
Sprint
Fastest way to test if we’re a fit. We ship one production automation against real data, with the same SLA every other build gets. You keep what we build either way.
- Visual editor
- Self-hosted
- LLM + API + code
- Scales out
Build
The default engagement. Multiple automations, the Systm portal, integrations and training. Phased so you can stop at any milestone.
- Local LLMs
- No per-token bills
- Sized to workload
- On your LAN
Audit
We review an existing system or build plan and deliver a written assessment. Often saves more than the fee.
- Local chat
- Trigger automations
- Reporting & runs
- Role-based access
§ 05 / WHAT WE DON'T DO
The short list.
Most software fails not because the code is wrong, but because nobody wanted to use it. Here’s how we avoid that.
No fixed-bid death marches
We work in two-week iterations with a target budget and visible burn-down. Stop any time.
No SaaS data egress
Inference runs on a box in your office. Documents, prompts and outputs never leave your network unless you decide they should.
No per-token AI bills
We size hardware to the workload once, not meter-by-meter forever. No usage surprises at the end of the quarter.
No black-box AI
If we use an LLM, the prompts, fine-tunes and evals live in your repo. No vendor magic.
No selling what you don’t need
If a Google Sheet would do the job, we’ll tell you to build a Google Sheet.
§ 06 / FAQ
Before you
ask.
The honest version. If your question isn’t here,
send it our way, we’ll answer it the same way.
01 Who is actually on my project?
A partner plus one or two senior engineers, depending on scope. No PMs, no account managers, no offshore team. The names on the contract are the people in the standups.
02 Do you do fixed-bid contracts?
Rarely, and only when the scope is genuinely fixed (e.g. an audit). Most automations are run as short iterations against a target budget — you see the burn-down weekly and can stop at the end of any phase.
03 How much does a typical engagement cost?
Costs depend largely on the scope of your needs, and we'll provide pricing based on your specific requirements.
04 Do you sign IP, NDAs and SOWs?
Yes - with an honest split. Your data, your configuration (prompts tuned to you, integrations to your tools, business rules and routing) and the running automation as deployed in your environment are yours to operate, perpetually. The underlying workflow IP — the reusable patterns and building blocks we’ve refined across many clients — stays with Systm; that library is the value you’re hiring us for. We never reuse your data, configs or branding. Your standard NDA is almost always fine.
05 What if you haven’t worked with the APIs in our stack?
There are hundreds of API endpoints across the modern SaaS world and we’ll be honest — we haven’t touched every one. Occasionally we hit an endpoint or a quirk of a system we haven’t worked with before, and a day or two goes into getting familiar with it. The experience we’ve accumulated across many clients makes that procedural, not risky — we know what to look for, what usually breaks, and how to write integrations that survive the next API version bump.
§ 07 / START
Bring us a process.
We'll bring the Systm.
A 30-minute call. No deck. We’ll listen, ask sharp questions
and tell you whether what you want is a two-week build
or a six-month build.