Systems built for AI.
Not retrofitted with it.
Orchestrate models, files, data and decisions as one system you can see and check. Start locally in your browser with nothing to stand up, then extend it headlessly and durably into production.
A useful result should not mean starting over next time.
You have worked out the instructions, found the right inputs and decided what a good result looks like. Keep it: save those steps as a process and bring it new inputs instead of rebuilding the sequence in a fresh chat.
What it is
A canvas where the steps around a model are connected: read a file, call a model, check a condition, write a result.
What it does
It turns a good one-off result into something you can hand new inputs to, with the instructions, sources and checks attached.
How it works
Steps run in the order the work happens, so a wrong result points at the step that produced it instead of one oversized prompt.
From a new brief to a draft someone can review.
Take proposal preparation. Start with the client brief and your approved information: extract the requirements, flag the missing details, draft the relevant sections and put the result in front of a reviewer. The same shape covers document review and structured research, with you deciding the instructions and where a person steps in.
Lay the steps out and keep them in view.
Arrange model calls, file reads, data operations, tools and decisions on a canvas, then connect the information each step needs. The saved process is called a network.
One step, one job.
A model chosen per step.
Your own code and tools where the work has to be exact.
Run it yourself, hand it to your team, or point it at a list.
The same network, used three ways. The logic stays in one place and keeps one owner.
Run it while you work.
Open the network, give it today’s input and watch each step report as it goes.
Give your team an app.
Put a form in front of it: someone fills in the task, starts the run and works with the result without opening the canvas.
Point it at a list.
When the task arrives as a list, run it as a batch and follow progress, reported model usage and estimated cost per item.
Apps and batch runs are on the Automate plan. A batch runs in an active NodeFox environment, not as an unattended cloud service. The 14-day trial requires a card.
Built for the person who owns the work, not only the person who owns the codebase.
If a task arrives regularly, has a right answer someone would recognize, and takes judgment somewhere in the middle, it fits. Pick the closest one.
You do the same work for every client.
Intake, review, drafting and delivery follow the same shape for every client, and you are the one holding it together. Build that shape once and spend your attention on what actually differs.
Start with proposal preparationRequests arrive faster than anyone can work through them.
A form, an email or a file arrives, and someone has to turn it into a usable record before anything else can move. The process does that preparation the same way every time; the judgment call stays with a person.
Start with operationsEvery document has to be checked before it leaves the building.
Contracts, submissions and reports get read against the same criteria every time. Turn the first pass into a process that produces a list of possible issues, with the decision still made by your reviewer.
Start with document reviewThe question repeats; the sources change.
Recurring research and publishing preparation follow a known route through known sources. Keep the route, change the inputs, and keep the editorial judgment where it belongs.
Start with research briefsWhere it fits: work that repeats, has a recognizable good answer, and passes a person before it counts. Where it doesn’t: a one-off question — that is a chat assistant — or a job that has to run untouched at three in the morning.
You add a process, not a platform.
NodeFox does not sit underneath your systems or ask you to move anything into it. It sits between them and takes on the part currently done by a person with fourteen tabs open. Setting up the steps is an afternoon’s work; after that, a morning of reading and retyping arrives as a draft with its sources attached.
Nothing to stand up.
Your systems stay where they are.
One process instead of one build per task.
Room to extend when the canvas is not enough.
Long, branching, stateful work needs more than a trigger.
Most of what a business repeats is not one event and one action: a case opens and stays open, and somebody has to approve a thing before it is allowed to happen. NodeFox is designed for long-running, recurring, stateful casework rather than app-to-app automation.
Routes you define, not paths it decides.
State that lasts as long as the job.
Parallel work that converges on purpose.
Side effects that fire only when you allow them.
Processes that compose.
A visual tool usually means a straight line and a black box; code means only engineers can change it. Here the person who understands the work builds it, and the result still behaves like something built on purpose.
Visible logic makes the process inspectable, not a model’s answer correct — test representative inputs and the cases you expect to fail.
Think this fits your business?
Tell us the task. We will tell you honestly whether NodeFox is the right tool for it.
Get the details by email
Leave your work email and we will send what a rollout involves: what to build first, what to connect, what to check.
Book a strategy session
Thirty minutes on one recurring task: what comes in, what has to come out, and who reviews it.
Ask a question first
Send the task in writing, with no confidential material, and we will reply with a straight answer about fit.
A few things to know before you begin.
Do I need to code?
Where does it run, and what leaves my browser?
Can it run when I am not watching?
Will the output still need review?
Start with the work you already repeat.
Pick one task. Build its steps. Keep the process for the next time it arrives.
A free account needs no card. NodeFox is in beta — test your process with representative inputs before relying on it for important work.