Two ways to work together
Contracted research and engineering
XS1 Automation researches, designs and builds an agent system around one of your workflows, starting with a single loop.
Research partnership
Joint work on open problems in agent systems, with scope, outputs and confidentiality agreed up front.
Step by step
Six steps, from first message to a running loop
Timelines are set during scoping, once we have seen the workflow. What you receive at each step, and what we need from you, does not change.
Start
Tell us about the workflow
Use the Work with XS1 Automation form to describe the work, the tools it lives in and who owns it today.
You receive
- A direct reply from XS1 with scoping questions, not a sales deck
From you: A short description of the workflow and what done would look like.
Scope
An initial conversation
We talk through the workflow, the constraints and the kind of engagement that fits.
You receive
- A proposed scope for the first loop
- The kind of engagement that fits
From you: Someone who knows the workflow well enough to answer how it runs today.
Phase 01 · Map
Map the workflow
We map the workflow with the people who run it: the tools, decisions, data, approvals and handoffs, and where it currently stalls.
You receive
- A map of the workflow as it runs today
- A review of constraints and risks
- A ranked shortlist of the loops worth automating first
From you: Time with the people who do the work, and a look at how it is done today.
Phase 02 · Design
Design the system
Agents, memory, permissions, integrations and reporting are specified before anything is built, including what may run on its own and what waits for approval.
You receive
- An architecture and routing plan
- A permission and approval model
- Success and rollback criteria
From you: Access decisions: which tools and data the system may touch, and who approves what.
Phase 03 · Deploy
Deploy the loop
The loop is built on private infrastructure, connected to your tools, and run against real work under supervision before anything runs unattended.
You receive
- A working loop connected to your tools
- Supervised first runs against real work
- Status reporting to the workflow owner
From you: A workflow owner who reviews the first runs and approves what the loop proposes.
Phase 04 · Improve
Improve the loop
The running loop stays instrumented. Blockers, exceptions and drift feed the next iteration, and scope grows only once the loop is stable.
You receive
- Progress and blocker reporting
- Iteration on real failure modes
- Scope expansion once the loop is stable
From you: Feedback on what the loop got wrong, and a decision on what it takes on next.
Deploy and Improve repeat for as long as the loop is in service.
Your side
What XS1 needs from you
Engagements move fastest when the operating reality is on the table early.
- 01
A named workflow owner
Someone who owns the workflow, approves sensitive actions and can pause the loop.
- 02
Access decisions
Which tools and data the system may touch, and at what level.
- 03
A definition of done
A concrete, checkable outcome for the first loop.
- 04
Real work to run against
Willingness to run the system on real work under supervision, not a sandbox forever.
After handover
Who runs the loop is agreed up front
Who operates and maintains the loop after handover (your team, XS1 or both) is agreed during Design and written into the engagement, along with how it is paused, changed or retired.
XS1 does not publish its team, partners or research. Qualified partners receive more detail after an initial conversation.
First stop
Ready to scope a first loop?
Tell us about the workflow, who owns it and what done looks like. We reply with scoping questions, not a sales deck.