Superscript × Candid Health

2026
Candid Health × Superscript
What an integration could look like

IN THE RULES ENGINE

•

00

Candid already runs the eligibility check this starts from

Candid pulls the patient’s coverage and holds the deductible, the copay and the position for the year. No eligibility response says what the visit will be billed as, or what this practice is actually paid for it. Without those two, patient balance cannot be set until the claim comes back.

What the eligibility response carries

01

We add one step before the visit

Our pricing runs alongside the rules Candid already runs, configured in the same place and by the same people. Where it sits in the stack is an integration detail, and there are several sensible answers to that. The one requirement is that it runs before the patient is seen.

The step we add

02

An eligibility response cannot produce this number on its own

The insurance card says $30. This visit is $142.02, because $350 of the deductible is still outstanding. The eligibility response carries the accumulator that accounts for the difference, but it cannot say what the visit will be billed as or what the practice is paid for it. Our four pricing engines complete the picture and return a price the patient can transact on, upfront.

One number, with the reason underneath it

03

We price what the visit may be billed as instead

Medicine is emergent. We price each alternative as a whole encounter at the patient's own cost share rather than as a difference from the base. Each one carries a likelihood taken from this practice's adjudicated claims.

Try an alternative

04

We price what may be added on the day

We price the whole cart rather than each code on its own, because benefits interact with each other. An add-on to the visit itself can fall under the same cost share, while a separate service carries one of its own, and a copay does not always cover what is added. Modelling how the benefits combine is what lets us price the variance instead of estimating it.

Switch an addition on

05

The price is ready before the patient is seen

The price is on the encounter days before it is coded, so it can be shown to the patient, collected, or carried into your own workflow ahead of the visit. Everything after that stays yours. We do not touch the claim, the submission or the adjudication.

What you get back

TRY A PLAN STATE

This is Candid Health's own Rules screen, rebuilt as a mock-up by Superscript. It is not built or endorsed by Candid Health.

GET /v1/prices/{price_id}200 · completed

{
  "id": "pr_6c40ae19",
  "status": "completed",
  "appointment": { "place_of_service": "02" },
  "patient_responsibility": 14202,
  "base": { "items": [ {
    "cpt": "99214",
    "label": "Telepsychiatry follow-up",
    "allowed": 14202,
    "adjustments": [ {
      "type": "deductible",
      "amount": 14202,
      "explanation": "$350.00 of the deductible
         was left, more than this line" } ] } ] },
  "x_axis": [
    { "cpt": "99215", "swaps": "99214",
      "relation": "severity", "likelihood": 0.07,
      "reason": "A higher-complexity visit",
      "patient_responsibility": 25293 },
    { "cpt": "99213", "swaps": "99214",
      "relation": "severity", "likelihood": 0.11,
      "reason": "A lower-complexity visit",
      "patient_responsibility": 10440 } ],
  "y_axis": [
    { "cpt": "G2211", "relation": "addition",
      "likelihood": 0.049, "billed_by": "provider",
      "reason": "Complex visit add-on",
      "patient_responsibility": 1610 },
    { "cpt": "90833", "relation": "addition",
      "likelihood": 0.033, "billed_by": "provider",
      "reason": "Psychotherapy with the E/M",
      "patient_responsibility": 8202 } ],
  "insurances": [ {
    "payer": "Aetna", "plan": "Choice POS II",
    "accumulators": {
      "deductible_remaining": 35000,
      "oop_remaining": 180000 } } ],
  "disclosure": "Exact, subject to our data inputs."
}

WHAT THE API RETURNS

•One call per encounter returns the patient's price, what the visit may be billed as instead, what may be added to it, and where the patient stands for the year.

patient_responsibilityThe patient balance no longer waits for adjudication. The figure your accounts receivable can only hold after the remit is available before the encounter.

base.items[]We price each predicted code on its own. A service line and its price agree before the claim exists.

adjustments[]Every amount says which benefit set it, in words. The eligibility response you already pull becomes a reason a patient can read.

x_axis[]We return what the visit may be billed as instead. We re-price each alternative in full and give it a likelihood from that practice's own claims.

y_axis[]We return what may be added on the day. We price each addition against the encounter as it stands, so it reads as what it actually costs.

insurances[].accumulatorsWe return where the patient stands for the year. The deductible and out-of-pocket positions the price was computed against come back with it.

THE PROTOCOL

•A Four-Engine Agentic Solution to Healthcare's Most Complex Data Problem.

We spent four years in R&D building Healthcare's First Pricing Protocol.

01AIM

APPOINTMENT

TYPEFollow-up · 30 min POS02 Telemedicine HISTORY99214 ×3

99214FOLLOW-UP VISIT88%

99213FOLLOW-UP, BRIEF11%

G2211COMPLEX ADD-ON5%

Predicts the treatments a visit will be billed as, before it happens, with the probability of each.

AIM (Appointment Inference Model)

Multilabel models trained on each practice's own adjudicated claims predict the CPTs a scheduled visit will be billed as, from more than 40 features about the patient, the appointment and the practice, tuned on three years of selecting treatments in the wild.

02ABE

ALL CLAIMS

99214 E/M 90833 Psych 99204 New pt G2211 Add-on

AETNA $142

CIGNA $162

ANTHEM $168

Generates payer-negotiated rates across procedures, providers and payers.

ABE (Algorithmic Balancing Engine)

Financial, actuarial models built on adjudication outcomes to resolve payer/plan bundling logic and CPT nuance. Over 1TB of data ETL per practice.

03AIR

271

$350 DED $30 COPAY ACTIVE

AETNA POS II

Patient X

An agentic benefits parsing engine to fully understand a patient's real-time coverage.

AIR (Accessing Insurance Resources)

Interprets X12 271 eligibility responses across EB segment hierarchies and free-text MSG fields using LLMs to normalize payer coverage logic complexity. Where a partner already runs eligibility, AIR reads the response they have rather than duplicating the call.

04ART

OUTPUTAETNA POS II

99214Follow-up visitDED$142.02

G2211Complex add-onDED$16.10

90833PsychotherapyCVRDCovered

PATIENT OWES$158.12

CLAIM DATA → ← MODEL UPDATE

A feedback loop to continuously tighten our pricing accuracy.

ART (Adjudicating Real-time Transactions)

Reads every 835 remittance and maps each pricing discrepancy. 8,700+ weekly adjustments generate labeled training examples with unambiguous outcome signals, so the system self-corrects without human annotation or reward model drift.

[ 05 ] WHO WE ARE

•Superscript builds pricing infrastructure for US healthcare, from New York, since 2021.

SOC 2 Type II · HIPAA business associate · Founded in New York in 2021. hello@superscript.nyc →

SOME OF THE GROUPS PRICING WITH US TODAY

+ MORE

OUR ADVISORS

Dr. Arthur KleinFormer President, Mt. Sinai Health Network

Jeffrey SachsFormer White House Advisor · Sachs Policy Group

Dr. Afsheen AfsharFormer Chief Data Officer, JPMorgan · Cerberus

Dr. Chris PittmanFounder & CEO, Vein911 · Chairman, HPS

Bob GlazerFormer CEO, ENT and Allergy Associates

Dr. Munish KhanejaCMO, Parachute Health · Former CSO, CareAbout

Joe BlewettFormer CEO, Kyruus (formerly Epion)

Sap SinhaFormer President and COO, Allied Digestive Health

SOME OF OUR INVESTORS

Mark Cuban Samsung Next Ventures M25 Health Lifepoint Health First Minute Capital Long Journey Ventures Lucy Guo Co-founder, Scale AI Arash Ferdowsi Co-founder, Dropbox and others

A WORLD-CLASS TEAM FROM Harvard, Meta, Yale, Apple, Stanford, Cigna, CNN, MIT, Amazon, Bridgewater, Google