How we work
Map the work. Then build the system.
Automation makes a process faster, broken parts included. So every engagement follows the same five steps, from sitting with the people who do the work to handing over a system they can run without us.
The five steps
Each step ends with something you keep.
Nothing is built before it is understood, and nothing goes live without monitoring. Here is what happens at each step, who takes part and what you have at the end of it.
Map
Sit with the people who do the work.
- What happens
- We walk through the work as it happens today, with the people who do it. Where requests come from, which tools they touch, what gets copied by hand, where things wait and where someone makes a judgment call. We look at real examples, not the process as it was written down.
- Who’s involved
- The people who do the work every day and the person who owns the outcome. From our side, the engineer who will build the system.
- What you get
- A written map of the process as it runs today: each step, who does it, which tool holds which record and where it gets stuck.
Design
Clear logic before code.
- What happens
- We decide how the new system should behave before building any of it. Which tool owns each record. What starts each run. What the system does on its own, where it waits for a person and what happens when a step fails.
- Who’s involved
- The person who owns the process signs off the design. We work through the edge cases with the people who know them best.
- What you get
- A design you can read without being an engineer: the flow, the rules, the checkpoints and the failure paths, agreed before the build starts.
Build
The right tool for each part.
- What happens
- Each part goes to the tool that fits it: a workflow platform where it does the job, custom code where it doesn’t. Rules handle what is predictable, and AI handles only what rules can’t. Every piece is tried against your real examples before it touches live data.
- Who’s involved
- Our engineers build. You see working pieces as they are ready, not one big reveal at the end.
- What you get
- A working system on your own accounts, tested with your own examples.
Run
Monitored from the first day.
- What happens
- The system goes live with monitoring already in place. Every run is logged. A failed step is retried when that is safe, and raised to a person when it isn’t. We watch the first weeks closely and adjust what the real work shows us.
- Who’s involved
- Your team uses it. We watch the logs and tune the rules and checkpoints with you.
- What you get
- A system in production, a log of every run and alerts that reach the right person.
Hand over
Yours to run, change or take elsewhere.
- What happens
- We hand over what is needed to run the system without us: the documentation, the access and a walkthrough with the people who will look after it.
- Who’s involved
- Whoever will own the system on your side, technical or not.
- What you get
- A system that runs on your accounts, with its code, workflows and documentation. Nothing depends on us to keep working.
Human checkpoints
Some steps should wait for a yes.
Automation handles the repetitive part. Judgment stays with your team. We agree together where a run stops and waits for a person, and you can move those points later.
Refund for order 5520
Needs your OK- Customer
- Jonas Berg
- Reason
- Item arrived damaged
- Why it stopped
- Above the limit for automatic refunds
Illustration · sample data
How a checkpoint works
- The run stops at the step and holds what it prepared.
- The right person is told, with what they need to decide.
- They approve it, edit it or skip it.
- The decision is logged with their name, and the run carries on.
Where they usually sit
Anything sent in your name
A reply to a customer, a quote, a message to a prospect.
Changes to records that matter
A deal stage, a contract, a customer’s details.
Money
A refund, a payment request or a purchase above a limit you set.
Anything the system is unsure about
A request it can’t sort, or data that doesn’t match what it expected.
Tools
The tool follows the process.
We don’t arrive with a platform to sell. Once the work is mapped, each part goes to the tool that fits it, and we write code where no tool does.
Workflow platforms
For workflows that fit them, and that your team may want to adjust without an engineer.
- n8n
- Make
- Zapier
Data
Where records live once a shared spreadsheet stops being enough.
- AirtableUsed in WTB Radar
- PostgreSQLUsed in Heart Capital Inc.
Software
For platforms, portals and anything a workflow tool can’t do well.
- Next.jsUsed in WTB Radar and Heart Capital Inc.
- OdooUsed in Heart Capital Inc.
- DockerUsed in Heart Capital Inc.
- Custom code
Models
For reading, sorting and drafting, behind rules that run first.
- OpenAIUsed in WTB Radar and Heart Capital Inc.
- Other model providers
Handover
Everything you need to run it without us.
A system is only yours if you can run it, change it and move it. So handover is a step with its own deliverables, not a last-day email.
Documentation
What each part does, what it touches and what to check when it stops. Written for the people who will look after it.
Your accounts
The system runs on accounts in your name, from the workflow platform to the model provider. Credentials and admin rights stay with you.
The code and the workflows
Repositories, exported workflows and configuration, in accounts you own and can reach without us.
A walkthrough
A live session with whoever will own the system: how it works, how to change it and time for their questions.
Monitoring that reports to you
Logs and alerts point at your team, so a problem reaches the person who can act on it.
From our work
For Heart Capital, handover was an eight-document pack covering every system we built.
Questions
Questions we’re asked first.
If yours isn’t here, bring it to the call.
What happens on the first call?
Thirty minutes on Google Meet. You walk us through the work that slows the business down, and we ask how it happens today, where it sticks and what it touches. If we can help, we say how we would approach it. If we can’t, we say that too.
Do you only use no-code tools?
No. We use n8n, Make or Zapier when they fit, and write code when they don’t. The tool follows the process, not the other way around.
Who owns what you build?
You do, once the work is paid for. Systems run on your accounts, and the code, workflows and documentation are handed over when the work is done. A product or venture partnership may set different terms in its signed agreement.
What happens when something breaks?
The systems we build log every run, retry what can safely be retried and raise an alert when a person needs to look. A failure is reported, not hidden.
Can a person approve steps before they run?
Yes. Any step can wait for a person: a reply to a customer, an update to a record, a payment request. You decide where those checkpoints sit, and you can move them later.
Do you work with businesses outside the US?
Yes. PlugWheel LLC is registered in Wyoming and works remotely. Our client work includes a recruitment platform for a firm serving Guyana, Trinidad and Suriname.
Start here
Step one is a conversation.
Mapping starts on a free 30-minute call. Bring the process that slows the business down, and we’ll start drawing it with you.
Free discovery call · 30 minutes · Google Meet