Most of what a business calls admin is a rule nobody wrote down.
We automate the repetitive middle of your operation — the coordinating, the chasing, the retyping — as workflows that execute your own policies, wired into the tools you already run on.
What it is
Almost every business has a process that works because a specific person remembers to do it. They read the request, decide whether it is allowed, update three systems, message the people affected, wait for replies, chase the ones who do not answer, and only then move the thing forward. It is slow, it is easy to get wrong, and it does not survive that person taking a holiday.
Automation here does not mean replacing judgement with a model. It means writing down the rules that were already being applied from memory, and letting them run the same way every time — with an AI step reserved for the calls a fixed script genuinely cannot make.
What we do
- End-to-end lifecycle workflows
- The whole path rather than one step of it: intake, policy evaluation, creation of the downstream work, changes, cancellations, refunds, credits and vouchers, each following the rules your business already has.
- Approval gathering that waits properly
- Workflows that ask several people at once, know who has answered and who has not, hold until the required responses are in, and reassign automatically when someone cannot take the work.
- Wired into the tools you already use
- Airtable, Google Calendar, Slack, SMS and email kept in agreement, so a change in one place is not a change somebody has to remember to make in four others.
- An AI step where judgement is genuinely needed
- Model calls reserved for the parts a rule cannot express — classifying a messy request, drafting a reply — rather than sprayed across a process that a condition would have handled more reliably.
How we work
- Your policies encoded as they are, not as a vendor's template assumes them to be
- Built on platforms your team can read and adjust: Make, Zapier, n8n
- Every automated decision traceable, so staff can see why something happened
- Deterministic rules first; a model only where a rule genuinely cannot decide
Built with
- Make
- Zapier
- n8n
- OpenAI API
- Retrieval-augmented search
- Transcription & topic tagging
- Elasticsearch
- Ingestion queues
- Airtable
- Fillout
- Softr
- Next.js
- TypeScript
- Node.js & Express
- Webhooks & scheduled jobs
Work delivered under this
Nida
Day-to-day operations moved onto automated flows across Make and Zapier, with an AI step for the judgment calls a fixed script can't make.
Read the story
Multi Compliance LMS
A SaaS platform companies use to stand up their own compliance LMS — their content, courses, pipelines and payments — with around 100,000 videos, documents and articles, all searchable.
Read the story
Property Cloud
A subscription CRM for real estate agencies. Listings, clients and the day-to-day admin sit in one place, with AI handling the repetitive parts.
Private project
Tell us what you need ai automations for.
Describe what’s slow, stuck or missing. You’ll leave the first conversation with a scope, a budget and a ship date — and if this is the wrong service for your problem, we’ll say so.