Skip to content
YC Root AccessYC Root Access

Building the Agentic AI Platform for Hospitals

Bunkerhill Health recently raised $55M to help build a true state-of-the-art AI platform for hospitals. In this episode of Founder Firesides, YC's Ankit Gupta sat down with their co-founder & CEO Nishith Khandwala to discuss how their tools dramatically speed up hospital operations, the cold email that landed them Cleveland Clinic as their first customer, and a future where even the most complicated surgeries are managed end-to-end by agents. https://www.bunkerhillhealth.com Chapters: 00:00 — $55M Series B Announcement 00:52 — What Bunker Hill Health Does 03:50 — Why It Takes Two Years to Onboard One AI Tool 05:09 — The Platform: Knowledge, Reasoning, Action 07:12 — The Innovator's Burnout Problem 09:57 — How Bunker Hill Actually Solves This 13:01 — How Nish Got Into Healthcare AI 17:00 — His Dad's Heart Attack Changed Everything 19:32 — How LLMs Transformed the Opportunity 22:05 — Cold-Emailing Cleveland Clinic 25:18 — Finding the Right Abstraction 28:04 — How Different Are Hospitals From Each Other? 31:12 — LLMs, Tool Use, and Hallucination 34:37 — Measuring Against the Standard of Care 39:27 — Building a Team of 21 43:30 — The Turkey Bone Patient Journey Apply to Y Combinator: https://www.ycombinator.com/apply Work at a startup: https://www.ycombinator.com/jobs

Ankit GuptahostNishith Khandwalaguest
Jul 16, 202647mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

Bunker Hill builds agentic AI platform to speed hospital iteration

  1. Bunker Hill Health announced a $55M Series B led by Khosla Ventures to build an agentic AI platform for health systems rather than isolated point solutions.
  2. Hospitals struggle to adopt AI because onboarding each new tool can take a year or two due to procurement, security/privacy, IT bandwidth, and change management bottlenecks.
  3. The company’s core product thesis is a reusable platform with three pillars—knowledge (connectors), reasoning (AI/LLMs), and action (workflow execution)—so solutions don’t stop at dashboards.
  4. LLMs changed the market by turning many “bespoke algorithms” into prompts and enabling tool-use plus citations, which reduces hallucination risk and expands feasible hospital workflows.
  5. Bunker Hill emphasizes evaluating AI against the existing “standard of care” (often imperfect or even ‘doing nothing’) and builds a lean team focused on platform primitives while forward-deployed staff configure many use cases per customer.

IDEAS WORTH REMEMBERING

5 ideas

The real bottleneck is deployment, not model capability.

Executives and clinicians often agree AI would help, but the time and organizational cost to onboard each new tool overwhelms the incremental value of any single point solution.

A hospital AI platform must both read and write to systems of record.

Bunker Hill argues “connectors in” (EHR/ERP/imaging/payer portals) are insufficient unless paired with “actions out” (messaging, scheduling, prior auth, billing) that execute workflows.

Knowledge–Reasoning–Action is the reusable abstraction across clinical and admin use cases.

The same pattern supports diverse workflows—incidental findings follow-up, registry abstraction, Medicaid eligibility identification, revenue cycle coding—by swapping inputs, prompts/tools, and actions.

Workflow variation—not data standards—is where customization is hardest.

FHIR/HL7 and imaging formats are relatively consistent across large systems, but each hospital’s processes and clinical preferences (e.g., Fleischner vs Brock scoring for lung nodules) require flexible “primitives” rather than rigid software.

Tool use plus citations is a practical approach to reducing hallucinations in clinical ops.

Instead of expecting the LLM to “know everything,” agents can fetch guidelines/payer policies as needed and must provide traceable links back to the originating note or record for human verification.

WORDS WORTH SAVING

5 quotes

We are really obsessed about lowering the cost of iteration in healthcare.

— Nishith Khandwala

You will spend two years trying to onboard a new tool, and only for it to deliver one specific thing. Imagine if for every app you ever used on your iPhone, you had to buy a separate phone.

— Nishith Khandwala

You build something, it goes nowhere, and suddenly you're just expected to show up for your next initiative without any loss of excitement. That's just not gonna work.

— Nishith Khandwala

If we don't take an automated action on top of that, you just have a dashboard that allows you to admire the problem as opposed to actually do something around it.

— Nishith Khandwala

I have been a strong believer that the performance comparison should not be 100%. It should always be against the standard of care.

— Nishith Khandwala

$55M Series B and investorsHospital AI adoption friction (procurement, security, IT, change management)Platform vs point solutionsKnowledge–Reasoning–Action architecture (Carebricks)Lowering cost of iteration and innovator burnoutHospital workflow variability vs standardized data formats (FHIR/HL7, imaging)LLM tool use, citations, and measuring against standard of careLean team model and forward-deployed deploymentEnd-to-end patient journey example (turkey bone → cardiac bypass)

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

iOSAndroidClaudeChrome