How it works

1. 0

   Step 0 · Discovery call (30 minutes, free).

   We talk about one AI tool you already use and what you need to know about it. By the end of the call, both of us know whether a review would help you. If it won't, we say so.

2. 1

   Step 1 · Agree on the need (week 1).

   Together we write down what you need from this review, who owns it, and what success means to you. Success is often what you were told the tool would do when you bought it, such as faster reviews at the same quality. We lock this in before we look at any data, so the answer can't be bent to fit the numbers.

3. 2

   Step 2 · Send your records (weeks 1 and 2).

   You export files from the systems you already have: decisions, review times, what the AI recommended, appeals. You send them to us securely. There's no software to install and no system integration. We never accept patient data without a signed business associate agreement.

4. 3

   Step 3 · Check the records and agree on the plan (week 2).

   We check every file and show you what's missing or can't be used. Then we write the analysis plan: what we will measure, how, and what we will compare it with. You and we agree on the plan before any analysis starts.

5. 4

   Step 4 · We analyze (weeks 3 to 5).

   Using data analytics and machine learning, we read what your people do, what your AI does, and what happens when the two work together. We compare AI-assisted work with your human-only work, case for case. Experienced reviewers re-check a random sample without seeing the AI's answer. We measure what it all costs you, and we tell you how sure we are of each finding.

6. 5

   Step 5 · Your report (week 6).

   You receive a report with your baseline, where you stand today, and your gaps, where your AI and your people fall short of the success you defined in step 1. Every number in it traces back to the records behind it.

Fixing what we found

6. 6

   Step 6 · Design the fix (agreed with you).

   We talk through your report together. For each gap, we work out the fix with your decision owner: the review rules, the training, and the changes to how information reaches your AI. You decide which gaps to fix first.

7. 7

   Step 7 · Build the fix (agreed with you).

   We put the fix in place with your people: checklists for the cases that need a person's review, training for each role that works with the AI, and written rules for when your AI vendor or model changes. Changes inside your AI tool are made by your IT team, your vendor or us, as we agree with you.

8. 8

   Step 8 · Measure again (agreed with you).

   We run the same checks on your new records and compare them with your baseline from step 5, so you can see what changed.

What we need from you: one decision owner, your data files, and time for a short working session at each step.