Candid Health's $120M: AI bolted to a painful workflow

The AI funding that closed this week wasn't for a chatbot or a foundation model. It was for medical billing. On July 21, Candid Health raised a $120M Series D led by Sixth Street Growth to automate revenue cycle management — the deeply unglamorous work of getting healthcare providers paid. The size of the round is less interesting than why it got funded, and that reason is a template.

What actually happened

Per Candid's announcement and FierceHealthcare's reporting, the round was led by Sixth Street Growth with Oak HC/FT, 8VC, and Y Combinator participating. The details that got it done:

• The raise tripled Candid's valuation over its Series C last year.
• 190% year-over-year annual contracted run-rate growth, and 180% net dollar retention in 2025.
• More than 200 healthcare organizations on the platform.
• The target: the roughly $280 billion spent every year on U.S. healthcare RCM — claims, denials, and the follow-up that eats staff time.

RCM is not a vision. It's a workflow: submit a claim, get it denied for a technicality, resubmit, chase the balance. Candid didn't sell "AI for healthcare." It sold "the thing that stops your claims from bleeding revenue," and it has the retention numbers to prove providers keep paying.

Why this matters for your business

The same-day funding roundup that carried Candid noted the common thread across every deal that closed: the winners "were not selling abstract AI promise. They were attached to a painful workflow, a regulated pathway, a procurement engine, or a capital allocation problem." That's the whole game, and it applies whether you're raising $120M or automating a $60K/year line item.

Here's how to copy it without a term sheet:

1. Name the workflow, not the technology. "AI for our business" is unfundable and unbuildable. "The AR follow-up that takes two people three days a month" is both.
2. Attach a number before you build. Candid can point to 180% net dollar retention. You should be able to point to hours saved or dollars recovered. If you can't measure it, you can't prove it worked — the same discipline behind automations that pay for themselves.
3. Automate the repetitive part, keep the judgment. RCM works for AI because most of it is rules, not decisions. The best targets in your shop are the same: high-volume, low-variance, deterministic where it counts.

Candid's $120M is a bet that boring, measurable automation beats general-purpose AI. We've been building on that bet the whole time.

Sources: Candid Health via BusinessWire, FierceHealthcare.