How we work
How an engagement works.
The partnership starts at discovery. Five stages run in one fixed order, each with a named owner: Develop is ours, and the other four are shared.
Five stages
-
01
Discover
We map the operation and the decisions it makes each day, with the people who own each one.
DeliverableA decision map: the decisions, the data behind each one, and an owner for each.
Owner: Both -
02
Design
We design the decision cards, rules and access levels with your team.
DeliverableA build plan that a named person signs off before work starts.
Owner: Both -
03
Develop
We build on our shared framework and test each change before it is released.
DeliverableTested changes, each checked against its acceptance criteria.
Owner: Ryca -
04
Deploy
We release in one fixed order: verify, deploy, a smoke test on each live client, then a person marks it shipped. New work stays off your system until it is ready.
DeliverableA shipped release. A failed smoke test blocks it.
Owner: Both -
05
Run
We operate and change the system with you after go-live. Changes come in as tickets and change requests.
DeliverableSupport tickets on business-hours clocks, and on request, a monthly update on what shipped and what is still open.
Owner: Both
Your Help Center
Your Help Center shows your tickets, release notes and the updates we have sent you.
Standard items are visible to everyone with Help Center access in your organization. On request means named approvers in your organization. Internal notes, build numbers and preview links never reach your view.
When we wait on you
Every open item has one owner, us or you, so both teams can see whose move it is.
- Marked
- The item shows that it waits on you until your reply comes in.
- Reminder
- After two business days, a reminder is drafted for a person to review and send.
- Cleared
- Your reply returns the item to the queue. A closed ticket clears its wait.
Our innovation hub
Our innovation hub is a space we set aside in our Chattanooga, Tennessee office. Clients come to it for collaboration events, training sessions and workshops, and we host them there.
Decision models
AI prepares each decision, and a named person makes it. Each model trains on your data against a feature set agreed with you. Each version is verified, then released only after you approve it. Models are an add-on.
- 01
Read
AI reads only the records the signed-in person can open, with that person's access and never wider.
- 02
Propose
It gathers the evidence and drafts a decision card: the record version, the reasons, the suggested choice and the alternatives.
- 03
Decide
Approve, adjust or decline. Adjust and Decline need a reason of 1 to 500 characters. A rule the person sets can approve above its threshold.
- 04
Record
The decision, the name, the reason and the record version go on the record. If the record changed first, the card refuses and offers a fresh one.
- SKU: 4 units on handlive
- Forecast: 9 units in 10 dayslive
- Supplier lead time: 6 dayslive
- Sales: +18% in 8 weekslive
- Stock record: v37 pinnedlive
Proposed: reorder 12 units
- R1Demand of 9 beats 4 on hand.
- R2Supplier lead time is 6 days.
Illustrative. Invented data.
Only a person can deploy or merge, send anything to a client, charge, refund or void a payment, approve a price or convert a lead. AI changes a client record only on a person's live request, with a stated reason.
Next step
Start with the decisions your team makes each day. We map them with you in the first stage.