Understand the workflow
AI Workflow Audit
Map one recurring process, locate friction and risks, then return ranked recommendations and written next steps. The designed service gives a team a bounded decision before it commits to a build.
Aran Labs Studio / Selected work / 2026
Tidy Engines
Creative and service teams spend hours preparing the work around their work. We shaped Tidy Engines’ service model and built its website to explain where agents could help, who owns each decision, and how to start with one workflow.
The business problem
Start where the work waits.
A producer assembles the same project folders. A studio gathers research before every proposal. A specialist checks a shortlist before it reaches a client. These recurring handoffs are the starting point for the service—not a reason to automate every decision.
Implement an agreed change
AI Agent Build
The proposed delivery model covers one workflow, including discovery, connections to agreed tools, testing, controls, documentation and handover. Software costs and ongoing support are scoped separately.
These are designed service offers. This project covers their positioning, website, illustrated workflow examples and enquiry experience.
02 / Interactive storytelling
Three routes.
One connected team.
The designed workflows distinguish human judgment, human direction and human review. Choose a route to follow the handoff, then pull back to see a connected team. This is an interactive explanation of the service model, not a running agent network.
Select a route, then reveal the network. Replay and motion controls are inside the demo.
Read the workflow description
Human judgment: Start → person → end. A person handles work that depends on judgment or relationships.
Human direction: Start → person → agent → end. A person sets the brief and boundaries; the agent carries out the agreed work.
Human taste: Start → agent → person → end. An agent prepares the first pass; a person reviews and refines it.
The network reveal: The view pulls back to show more connected teams. A dotted “Where we started” frame identifies the original three routes.
A concrete first journey
From a recurring task
to a useful brief.
- Recognize the handoff
A studio lead explores a brief-to-proposal example: a person sets direction, an agent prepares source-linked material, and the team shapes the proposal. This workflow is designed, not connected to client tools.
- Choose a starting point
The audit and build pages explain scope, deliverables, exclusions and questions to resolve. Separate pages make it possible to discuss one bounded service rather than an undefined AI transformation.
- Prepare an enquiry
The working contact helper carries the chosen interest into a browser form, validates the brief and prepares an email draft. The visitor reviews it before opening their email app or copying it. The website does not send or store the enquiry.
A draft arrives at the reviewer. Rough marks become a star as the hands move, then the finished document continues.
Original handoff detail, enlarged. Plays once; replay to watch again.
03 / The human checkpoint
The last touch
belongs to a person.
The document pauses beneath the reviewer’s hands. Rough marks resolve into a star, with a small movement of the hands marking the change. Human refinement becomes a visible event in the workflow.
04 / Illustration system
A softer face
for systems work.
Woven surfaces make the robots feel tactile. Hair, glasses and clothing give people individual presence. Fine lavender, sage and terracotta lines connect them without overpowering the characters.
05 / Across screen sizes
The same story.
A different route down.
Desktop paths run across the page. Mobile rebuilds them as three vertical branches. Motion can be paused; reduced-motion preferences open a complete, still network, with every route available as a static view.
What we delivered
A service people can understand.
A website they can use.
Make ownership visible
Distinct people and woven agents give each handoff an owner. A pause beneath the reviewer’s hands marks a decision, rather than presenting every step as automatic. The drawings explain how the service is intended to work.
Keep the enquiry in the browser
The contact helper prepares text locally and exposes a reviewable draft. There is no submission endpoint behind it. That matches the current business stage and avoids collecting workflow details into an unfinished system.
Build a small, readable site
A static build produces the offer, example, contact and information pages with local fonts and artwork. The workflow redraws for narrow screens, supports pause and reduced motion, and has a text description. Visitors can understand the service without following animation.
Aran Labs delivered service positioning, offer pages, designed workflow examples, illustration, interactive motion, responsive layouts and the enquiry-draft helper.