Fix the order. Not the denial.
Every denial started as an order somebody could still have fixed. Revenue Guard checks each one against your payer rules before billing — while it is an edit, not an appeal.
Your rules. Every order. Before billing.
Start with what your billing team already knows — write the rules straight in. Have the payer’s policy document? AI reads it and drafts rules from it too. No rule goes live until your team promotes it. After that, Revenue Guard learns from the flags your team disagrees with, and some corrections apply themselves once a check against your own recent orders backs them.
- 1Written or extractedLAB-0231 Thyroid testing.pdf · p. 14or typed in by your billing team
- 2Drafted rule AI draftsTSH needs a thyroid diagnosisCascade Mutual Health · checks nothing yetDraft
- 3PromotedRosa Lindqvist, billing leadread against p. 14, then made liveLive
- 4Order checked AI checksREQ 4471-B · CMP, TSH reflexWhitfield, Dana R. · Z00.00, R53.8Risk
By the time billing sees it, the damage is done.
The order goes out. The claim goes out. Six weeks later the denial arrives — and now it is an appeal, a rework, or a write-off, for something that was a thirty-second edit at order entry.
| Svc date | Proc | Units | Billed | Allowed | Paid | Adj |
|---|---|---|---|---|---|---|
| 08/12/26 | 80053 | 1 | 42.00 | 14.10 | 14.10 | CO-45 |
| 08/12/26 | 84443 | 1 | 38.00 | 0.00 | 0.00 | CO-11 |
| 08/12/26 | 36415 | 1 | 6.00 | 3.00 | 3.00 | CO-45 |
CO-11The diagnosis is inconsistent with the procedure.
CO-45Charge exceeds fee schedule or maximum allowable.
Premier Inc. — 280 hospitals, 23 states, 2023 claims. Hospital figures, not lab. Not ours.
A rule from a document cites the line it came from.
Where AI drafted a rule, it never asks you to take its word for it — click through to the exact sentence, verified against the page before it was stored. Rules your team wrote say so plainly instead.
3. Coverage criteria
3.1 Tests in this policy are covered when ordered by the treating provider and billed with a diagnosis listed in Appendix A.
3.2 TSH (CPT 84443) is medically necessary for members with thyroid disease (E00–E07) or abnormal thyroid function findings (R94.6). TSH ordered as part of a routine examination (Z00.00, Z00.01) is not covered.
3.3 Free T4 (CPT 84439) is covered as a reflex when the TSH result falls outside the reference range.
3.4 Thyroid antibody testing (CPT 86376, 86800) is covered for members with a confirmed diagnosis of hypothyroidism or…
- When
- CPT 84443 and no ICD-10 in E00–E07 or R94.6
- Then
- Risk — not covered for this diagnosis
Three signals. One clear decision.
AI checks each order against the rules your team promoted. Every order comes back graded — Severe, Warning, or clear — and every finding names the rule that fired, why it fired, and the fix.
It grades. Your team knows what to fix now and what can wait.
84443 is billed with Z00.00 and R53.8. Policy LAB-0231 §3.2 (p. 14) covers it for E00–E07 or R94.6.
FixAsk Northgate Clinic for the diagnosis behind the TSH.
The payer needs the full code: R53.81, R53.82 or R53.83.
FixConfirm the full code with the ordering provider.
If the TSH runs uncovered, this plan requires a signed member waiver first.
FixCollect it at the draw if the diagnosis stands.
Nothing changes for your team.
Orders reach Revenue Guard from whatever you already run — LIMS, EHR, referral pipelines — over a single endpoint your systems team points at once. Nobody changes how they book an order, and no one logs into a second system to work a queue.
Send from any source
LIMS, EHR, RIS, referral pipelines — if it can POST JSON, it can send orders.
Only the essentials
Demographics, insurance, tests, ICD and CPT. No full chart, no heavy integration.
Signed webhooks
Every delivery is signed HMAC-SHA256, and the secret can be rotated.
# an order, from any source { "order_id": "ORD-10428", "payer_ids": ["b7c1e0a2-..."], "patient_info": { "patient_uid": "PT-5512", "dob": "1962-03-11", "sex": "F" }, "insurance_info": { "insurance_name": "Example Health Plan", "policy_number": "..." }, "ordered_tests": [ { "short_name": "HbA1c", "test_name": "Hemoglobin A1c", "test_code": "GLU-A1C", "cpt_codes": ["83036"], "icd_codes": ["E11.9"] } ] } → 202 { "orderId": "ord_9fQ2xK...", "externalOrderId": "ORD-10428", "version": 1, "orderVersionId": "ov_2Lw7tR...", "status": "created", "environment": "sandbox", "livemode": false }
- Every order, checked against your payer rules — before billing.
- Runs alongside your eligibility and prior-auth systems.
- Nothing to rip out — your LIS and billing stack stay exactly as they are.
Whoever reviews or ignores a flagged order is recorded against the version they saw, and ignoring a risk takes a written reason. How orders and policies are handled →
One plan. Every order checked.
A flat monthly fee covers your first 10,000 order checks each month. Past that, each check is 5¢.
Includes 10,000 order checks a month. 5¢ per check after that.
- Every order checked against the rules your team promoted.
- Unlimited people. Nobody pays per seat.
- API access, webhooks and recommendations included.
14 days free, up to 500 order checks, no card.
Paper becomes orders. Also with AI.
Faxed, scanned, handwritten — AiReq reads the requisition and turns it into a structured order your team approves. Same idea as Revenue Guard, pointed at the requisition instead of the order. Those orders are exactly what Revenue Guard then checks.
See how AiReq works →How it’s kept safe, how it connects, what feeds it
Point it at the payer you argue with most.
Fourteen days and 500 order checks, no card. Start with a single payer — the one your team argues with most — and see what it catches on orders you have already sent.
Revenue Guard is built by CrelioHealth and billed separately from the CrelioHealth LIS. Sign in