How it works
From your first message to a running system
Each step ends with something you can check: a written answer, a prototype on your data, a system accepted against an agreed test set.
Your message
Describe the workflow, the data and what a good result looks like. A reply comes from the person who would do the work.
A 30-minute call
The workflow, the data, the constraints (PHI, model risk, IT policy) and whether a Discovery Sprint would answer the open questions.
Discovery Sprint
Week one: workflow and data audit, a sample of the data, the evaluation set. Week two: a working prototype on that sample and a written plan with scope, risks and a quote.
Build
Milestones from the plan, each demonstrated on your data. Acceptance against the evaluation set agreed in discovery. Half on start, half on acceptance.
Operate
Monitoring of uptime, accuracy and cost; changes to prompts, data and interface; a monthly report with usage, quality and what changed.
Leaving
An export of your data, the code, the prompts and the documentation, and a handover call. Nothing stays locked with DF1.
Your part and ours
| You | DF1 |
|---|---|
| Describe the workflow and who does it today | Ask the questions that find the hard cases |
| Provide a data sample, or approve synthetic data | Audit the data and build the evaluation set |
| Name a reviewer for the prototype | Build the prototype and the written plan |
| Accept milestones against the evaluation set | Build, test, document and deploy |
| Use the system and report what feels wrong | Monitor, maintain and report monthly |
Tell us the workflow. We show you the system.
Describe the process and the data. A reply comes within one business day, from the person who would do the work.