Skip to main content

Give your operations team its own engineering function.

We work alongside your team to turn messy, manual workflows into production agent systems — starting with one live result in 30 days.

Who reads this

Fast-Growing Startups

Every engineer is busy on the core product. We handle the operations backlog.

Established Companies

Finance, legal, or logistics teams with little or no internal engineering.

The same job, before and after

Today

  1. Email arrives
  2. Someone copies it into a spreadsheet
  3. Someone checks it against the CRM
  4. Someone chases it on Slack
  5. Someone updates the CRM
  6. Someone flags it for the manager
  7. Manager approves

With OpenOnion

Email · CRM · Documents

One agent system

Classifies · updates · chases · records

Manager approves

Your team sees exceptions and decisions. Nothing else.

Your team keeps the judgement. The system handles the movement.

What We Build

Five places we have made that change. Each one is a system that runs, not a template you configure.

  • Recruitment Agents

    We find the people, reach them first, and hand your interviewers a full picture.

    LinkedIn, GitHub and papers feed a single screening step, which produces outbound contact and scheduled interviews.LinkedInGitHubPapersOutboundbefore they applyInterviews
  • Sales Agents

    Leads collected, contacted cold, and kept warm without anyone remembering to.

    Running now for Black Diamondz, luxury real estate in Sydney.

    Collected leads are enriched and de-duplicated, sequenced into cold outreach, and engagement is tracked over time.Lead 1Lead 2Lead 3Lead 4ColdoutreachEngagement
  • Integration Agents

    One change lands everywhere it needs to, without anyone copying between tabs.

    A change in one system is picked up centrally and pushed into the CRM, the finance system and the document store.Changein one toolCRMFinanceDocs
  • Contract & Ledger Agents

    Every agreement captured, dated and watched, so no renewal is a surprise.

    Scattered contracts are captured into a single register with dates, values and an early renewal flag.Contract registerrenewal
  • SOP & Wiki Agents

    The wiki stops being a graveyard, because something keeps it honest.

    How the work is actually done is compared against the written page; the stale section is flagged and a correction is published.How it's doneWhat's writtenupdated

How It Starts

  1. 1

    Book a call

    We align on your operational bottlenecks.

  2. 2

    We watch

    We sit with your team for a week and watch the job get done.

  3. 3

    We build

    The system is shaped around how the job is actually done, not around a spec.

  4. 4

    We run it

    Live within 30 days, and we keep it running while the next one starts.

The Terms

Start with one month
A subscription. Paid monthly, not per project.
One week to change your mind
Not working out? 75% back, no questions.
No lock-in
Cancel or pause any month.
Immediate ownership
You own what we build from day one.

Pricing

One engineer, on subscription.

$6,000

USD per month

One price. However many jobs you put in the queue.

It does not move with how many workflows you bring us, and there is no hourly meter to watch. Jobs are worked one at a time and reordered any Monday — the backlog is yours to prioritise, not ours to quote on.

  • No setup fee
  • No hourly billing
  • Cancel any month

Or pay us out of the increase

$1,000

USD per month

Plus a share of the increase, for 12 months.

$12,000 at worst. $144,000 at most.

Short-term rentals qualify. Most businesses don’t.

It needs a result that can be counted against a baseline we agree beforehand. Whether yours can is the first thing we will tell you.

After the 12 months it becomes the flat $6,000 — the same price as if you had never taken this route.

We agree the share before we start, never after we see your results.

Prices in US dollars. We invoice internationally.

AI shouldn’t just make each employee faster. It should make the whole company run differently.

Questions

How is this different from buying Zapier or n8n?

It is not a different tool — we use those where they fit. The difference is who runs them. A platform gives your ops lead a canvas and a login, and the maintenance becomes their second job: a form field changes, a vendor updates an API, and something stops silently until somebody notices weeks later. What you are paying for here is the person, not the canvas. If your team has the capacity to build and maintain this themselves, buy the platform — we will tell you so on the call.

We have engineers. Why would we not just ask them?

Ask them. If internal automation actually makes it into a sprint, you do not need us. In most companies it does not, and everyone involved knows why: it competes against work a customer is waiting for, and it loses correctly every time. We are not better engineers than yours. We are engineers whose entire quarter is this list.

You would need access to our systems. How does that pass a security review?

We take the narrowest access that does the job, in your own tenancy, under accounts you create and can revoke from your side without asking us. What those accounts can see and do is written down before you grant them, and that document is what your reviewer reads. Where your policy needs a human in the loop for a class of action, that becomes an approval rule rather than a promise.

Who owns what you build?

You do, from day one — not on exit and not after a term. The software underneath is our own open source under Apache-2.0, so there is no licence to buy back and you do not need our permission to bring in somebody else.

Can we stop?

Any month, and you keep everything we built. There is no notice period and nothing to cancel by phone.