01
Choose the bank, billing, and remittance files for the month.
Medical practices and billing teams
Find the claims that didn’t get paid right.
Reconciliator handles the payment-posting grind: linking insurance payments to your claims, chasing down denials, flagging claims paid less than your contract says, and checking patient coverage. Every finding links to its source records, and AI steps run through your practice’s own Azure OpenAI deployment.
Product model
$995 per seat each year. Azure OpenAI usage is billed separately through your own Microsoft account.
Rollout shape
A 90-minute workflow discovery and coaching session is recommended for most teams.
localhost:4876
Start here
Payment exception finder
Compare insurance payment files, billing reports, and bank deposits, then show every mismatch next to the records behind it.
02
Match insurance payments to claims and flag what does not add up.
03
Give staff a short list of items to review, with the records behind each one.
Proof demo
Watch Boxwood match a month of insurance payments to claims and flag what doesn’t add up.
Boxwood compares your insurance payment files, billing report, and bank deposits, then shows every mismatch next to the exact records behind it.
Read the video transcript
Medical billing reconciliation should already be automated. Comparing 835 files, practice exports, and bank deposits is repetitive. AI can do the comparison. The problem is the data. Patient records and payment details cannot be dropped into an ordinary AI tool. Legal and compliance say no. Boxwood turns no into yes. Reconciliator works from the billing records you choose. Source files, artifacts, and work history stay in the local workbench. When AI is needed, approved inference context goes through your organization's Azure OpenAI account. Use Reconciliator to compare remittance files, billing exports, and deposits. It flags mismatches and links each finding to its source, so you can inspect the evidence yourself. Billing work that ordinary AI could not be approved to handle can now be automated.
demo.mp4
More than one workflow
The first job proves it works. Then give Boxwood the rest.
Describe another job, choose the files and tools it can use, and run it. When the process works, save it so you or your team can use it again.
01
Billing reconciliation
Match payment files, bank deposits, and practice records, then send staff straight to the exceptions.
02
EOB and 835 handling
Stage remittance files, exports, and supporting records with explicit review points.
03
Denial follow-up packets
Pull together the denial reason, supporting records, missing-document checklist, and deadline for staff review.
How the workflow moves
From a pile of billing records to a short list of problems.
Choose the remittance files, billing report, and deposits. Boxwood flags what does not add up and shows the claim and payment behind each item.
01
Practice billing records
Stage approved remittance files, practice exports, and deposits in the local workbench.
02
Only what the task needs
The AI sees only the information you choose for this job through your organization’s private Microsoft account.
03
Reconciliator runs the workflow
The agent performs the repeatable comparison or review inside the workflow you defined.
04
A result you can check
The result returns to the workbench with the supporting records available to inspect.
Workflow coaching
You do not need to know what to automate before the call.
Unless you already build agents yourself, the coaching is where much of the value comes from. We help you see which parts of the day can become repeatable Boxwood workflows.
01
Walk through how your team actually spends its day.
02
Find the repetitive work Boxwood can handle and your team can check.
03
Choose the best first workflow and set it up together.
For legal, IT, and compliance review
Files stay stored on your machine. AI runs through your own Azure account.
Boxwood keeps source files, results, and work history on your computer. When a task needs AI, it sends that task's content to the Azure OpenAI deployment your organization creates and controls. Microsoft bills you for it directly; Ironwood never needs your key.
How HIPAA compliance applies to an AI agent ->01
Files are stored locally
Source files, artifacts, credentials, and work history live in the local workbench on your machine.
02
AI runs on your Azure account
When a step needs AI, its content goes to your organization's configured Azure OpenAI deployment. Not to a public chatbot, and not through our servers.
03
Every finding shows its source
Results link to the file, row, or record they came from, so the person reviewing can open the evidence directly.
04
Access is scoped
An agent can touch only the files and connectors you selected for that workflow.
Pilot inquiries
Bring one task your team already knows how to check.
Research a case, review documents, compare records, update a spreadsheet, or build a reusable internal tool. We will run your example through Boxwood and show you the result, the source trail, and where the information went.