ServiceOperations & hosting
Workflow automation: the repetitive tasks done by software, with a person still in charge
Our workflow automation service takes the tasks your team repeats every day, such as passing a new lead to the right person, delivering a product after payment, sorting an inbox or compiling a report, and makes software do them reliably, with AI where it helps. Every automation we build has a usage cap, a dry-run mode, an off switch and a log of what it did, and nothing that leaves your company (an email, a message, a payment) goes out without the rules you approved.
What we do for you
- A short workshop to list repetitive tasks and pick the few worth automating first
- Lead routing: form or chat submissions sent to one central place, with an instant alert and a record in your CRM
- After-payment automation: confirmation, delivery of a digital product by email, records updated
- Inbox sorting with AI: messages moved and labelled into folders, never deleted
- Scheduled reports and reminders sent to the right people
- Connections between the tools you already use, through their official APIs
- Guards on every automation: daily cap, dry-run mode, off switch, human handover, action log
- Documentation of each automation: trigger, steps, owner, how to stop it
Who it is for
- Fits: small teams who copy data between tools, chase the same reminders, or lose leads in an inbox
- Fits: online businesses selling digital products that should be delivered the minute payment clears
- Fits: owners who want automation but are wary of a bot acting on its own
- Does not fit: anyone wanting to send unsolicited messages at scale; we do not build that
Automations we already run
These run every day on our own sites and tools, which is how we learned where automations break:
- Lead capture: one central lead API receives the leads of all our sites, sends an alert in under 2 minutes, and files each lead in a CRM.
- Payment to delivery: one-click Stripe checkout, then instant delivery of the product by email.
- Email triage: an AI sorts our founder's mailbox every 2 hours, moving and labelling messages, never deleting any.
- Publishing: when our SEO engine ships a page change, sitemaps, Search Console and IndexNow are updated the same day.
- Team bots: our team's work bot on chat and WhatsApp files tasks and briefs into our internal system.
The guards on every automation
| Guard | What it does |
|---|---|
| Daily cap | Limits how many actions the automation can take per day, so a bug cannot run away |
| Dry-run mode | The automation shows what it would do, without doing it; used for every new rule |
| Off switch | One simple way to stop it, documented and known by your team |
| Human handover | When the case is unclear, a person decides |
| Action log | Every action recorded, so you can see what happened and why |
| No deletion by default | Automations move, label or flag; deleting data is a human decision |
Platform rules we build around
Automations that touch customers have to follow the platforms' rules. Two examples. On WhatsApp, Meta requires opt-in before a business messages someone: the business must clearly state that the person is opting in to receive messages and the name of the business, and must comply with applicable law. For payments, Stripe states that you cannot rely only on the page shown after checkout to fulfil an order, because a customer may pay and lose their connection before that page loads; Stripe recommends webhooks so that fulfilment happens for every payment, and asks that fulfilment run only once per payment even if it is triggered several times.
How we work
- List the repetitive tasks with the people who do them, and estimate the time each one takes.
- Pick two or three where the rules are clear and a mistake is cheap to fix.
- Build in dry-run first, and show you what the automation would have done on real cases.
- Switch on with your OK, with caps and logs, and review the first weeks together.
- Extend only once the first ones run cleanly.
Where AI helps, and where it does not
AI is useful for reading and sorting: classifying an email, extracting the details of a request, drafting a reply for a person to check. It is a poor choice for anything that must be exactly right every time, like a price, a tax rate or a payment; those stay in plain, tested rules. And anything that leaves your company in your name stays within the rules you approved, with a person able to step in.
Good first candidates
Poor first candidates are tasks that depend on judgement, on data scattered across people's heads, or on a tool with no reliable way to connect. We would rather automate two small things that run cleanly than one large thing that needs constant fixing.
- A task done the same way every time, by several people, several times a week.
- Clear inputs: a form, an email with a known format, a payment, a calendar event.
- A cheap mistake: if the automation gets it wrong, someone can fix it in a minute.
- A visible result: a lead answered faster, an order delivered at once, a report on Monday morning.
Questions we get
Will an automation send emails or messages on its own?
Only within rules you have approved, with a daily cap and a log. New automations start in dry-run mode, and anything unclear is handed to a person.
Can an automation delete my data?
Not by default. Our automations move, label or flag; deletion stays a human decision. Our own email triage only moves and labels messages; it never deletes them.
What tools do you connect?
The ones you already use, through their official APIs where they exist: email, CRM, payment platform, chat, spreadsheets. We tell you upfront when a tool has no reliable way to connect.
What happens if something goes wrong?
The off switch stops the automation, the log shows what it did, and the daily cap limits how far a mistake can go. We fix the rule, test it in dry-run again, then switch back on.
Sources
- WhatsApp Business Platform: Get opt-in for WhatsApp (checked 2026-10-06)
- Stripe Docs: Fulfil orders (Checkout) (checked 2026-10-06)