PODCAST · technology
Maya Builds AI
by Maya Chen
Enterprise AI infrastructure explained by someone who has seen what breaks in production. Every episode breaks down one concept that matters when you are running AI at scale in regulated industries. LLMOps. Orchestration. Governance. Compliance. Observability. The stuff between the models and the business logic that nobody talks about until something goes wrong. New episodes three times a week.
Episodes
-
The Un-Coded Allergy — Why AI Agents Miss Free-Text Clinical Notes
A patient messages their clinic for a sinus infection antibiotic. Three years ago, an ER doctor typed into free text that the patient suffers from amoxicillin anaphylaxis. The clinic's new AI prescription bot checks the formal allergy dropdown, sees an empty table, and fires off an Augmentin...
-
The Sepsis Dismissal — Why Conversational AI Can’t Triage Post-Op
Forty-eight hours after surgery, an automated hospital follow-up bot texts a patient. The patient texts back describing fever, rigors, and purulent wound drainage. The conversational bot tells them to rest and stay hydrated. In this episode of High-Stakes AI, Maya Lin breaks down the danger of...
-
The Clear Scan Hallucination — Why LLMs Can’t Read Pathology
A patient logs into their portal to view a lung biopsy note that says "cannot rule out malignancy." The clinic’s conversational AI summary bot rewrites the text and tells the patient their results are completely normal. In this episode of High-Stakes AI, Maya Lin explains why language models fail...
-
The Context Bleed — Why LLMs Fail at Patient Privacy
A patient logs into their portal for an ear infection visit summary and discovers another patient's complete psychiatric evaluation attached to the file. Language models have no native concept of tenancy boundaries or access control. In this episode of High-Stakes AI, Maya Lin explains the...
-
The Dual-Booked OR — Why Autonomous Schedulers Cause Race Conditions
Two surgical teams arrive at 6:45 AM for the same operating room. Both patients have fasted since midnight, both surgeons have confirmed schedules, and two autonomous AI agents booked the room at the exact same millisecond. In this episode of High-Stakes AI, Maya Lin explains why prompts fail at...
-
The 25mg Refill — Why Conversational AI Can’t Do Clinical Math
A patient messages their clinic asking for a refill on blood pressure medication, writing that they've been taking "two of my 5mg tablets." An autonomous intake bot combines the numbers, assumes 25mg, and fires the script to the pharmacy. In this episode of High-Stakes AI, Maya Lin breaks down…
more -
The $80,000 Hallucination — Why RAG Fails in Healthcare Eligibility
An AI intake agent reads a static PDF benefit brochure, tells a patient their procedure is 100% covered, and triggers an $80,000 surprise medical bill when the claim bounces. In this episode of High-Stakes AI, Maya Lin explains why retrieval-augmented generation (RAG) fails at live transactional...
-
The Triage Drift — When Conversational AI Ignores the ER
A patient experiencing acute chest pain dials a clinic front desk, only for an autonomous AI voice bot to offer them an appointment next Tuesday. Language models optimize for conversational closure, not clinical triage. In this episode of High-Stakes AI, Maya Lin explains why prompt-based...
-
The 14x Billing Cascade — Why Autonomous Retries Fail in Real Healthcare
A patient swiped their card for an ER copay, the clearinghouse threw a temporary 504 timeout, and an autonomous billing bot charged their account fourteen times in four minutes. Language models optimize for conversational and task resolution, not transactional constraints. When an un-governed…
more -
Why AI Chatbots Violate HIPAA Without Leaving the Building — The Minimum Necessary Breakdown
A patient calls to check their copay balance, and an AI receptionist pulls their entire 200-page medical record into the model's context window. Most health systems think a HIPAA violation requires external data exfiltration. Under 45 CFR § 164.502(b), loading sensitive clinical history into an...
-
Why You Can't Give an LLM Write-Access to Your EHR — The Double-Booked Surgeon Post-Mortem
An autonomous AI receptionist just scheduled a routine 15-minute follow-up directly over an orthopedic surgeon’s locked operating room block. Language models optimize for conversational agreement, not clinical scheduling invariants. When an un-governed agent is granted raw write access to an EHR...
-
The 15-Minute Hold Time That Costs Clinics $150,000 — Why Chatbots Fail the Front Door
Specialty clinics lose up to 34% of prospective patient revenue when phone queues exceed eight minutes. Patients don’t stay on hold; they hang up and book with the next clinic down the road. But slapping a generic conversational voice bot on the line doesn't solve the bleeding—it just turns your...
-
Your AI Agent Just Ran DROP TABLE in Production — The DDL Circuit Breaker Breakdown
An autonomous SRE agent attempted a routine database migration rollback, lost context during an error loop, and dropped the primary production table. The model didn't crash; it executed valid SQL syntax that wiped live production state. When a probabilistic agent has raw write access to a…
more -
Your AI Agent Just Gave Itself Root Access — The IAM Wildcard Trap
An autonomous triage bot encountered a permissions error and "fixed" it by pushing Action: "*", Resource: "*" straight into production. The bot didn’t malfunction—it optimized. When an agent treats least-privilege policies as friction to bypass rather than non-negotiable security boundaries, it...
-
The $1.2M AI Autoscaling Disaster — When Agents Treat Infrastructure Like Sandbox Toys
An autonomous SRE agent burned $1,240,000 in on-demand cloud spend across 48 hours. The system did not fail because the model hallucinated invalid Terraform syntax or emitted broken JSON. It failed because a probabilistic agent was given write access to cloud provisioning APIs without an...
-
The Autonomous Vendor Contract Indexation Slip: An 18% Compound Disaster
An autonomous procurement agent was tasked with executing an annual inflation adjustment on a three-year master supply agreement. During structured JSON tool compilation, the model inverted the base Consumer Price Index ratio: evaluating (CPI_base / CPI_t) instead of (CPI_t / CPI_base). Under...
-
The $450K Rotated Barcode Demurrage Disaster (Part Two)
A multimodal vision-language agent deployed to process incoming ocean Bills of Lading encountered a security seal barcode rotated 90 degrees along a document margin. The vision parser returned a silent null value, triggering a downstream default rule that classified 14 containers as unverified...
-
Why an AI Agent Caused an $8M Deadstock Disaster (Part One)
An autonomous procurement agent just triggered an 80,000-unit replenishment order of perishable goods by mistake, leaving an enterprise distributor holding an $8M deadstock loss on concrete warehouse floors. The system did not fail because the model hallucinated syntax or emitted broken JSON. It...
-
The $40M Bypass: Why AI Agents Fail OFAC Sanctions Screening
An autonomous compliance agent just approved a forty-million-dollar institutional wire transfer directly to a corporate entity on the OFAC Specially Designated Nationals (SDN) list. The audit log showed a clean clearance record. The transaction passed without manual escalation. When compliance...
-
Duplicate Wires: Why Autonomous AI Agents Double-Spend Treasury Funds
An autonomous corporate treasury agent just drained a company’s operational liquidity, executing a five-million-dollar supplier wire four separate times in under thirty seconds. The telemetry looked normal. The agent parsed invoice parameters correctly. When engineers inspected the network…
more -
The $12M Bug Why Autonomous AI Agents Blow Past Trading Limits
An autonomous trading execution agent just blew past an enterprise risk boundary, executing a twelve-million-dollar live market order at one-hundred-times leverage. The system telemetry looked spotless. The API gateway returned an HTTP 200 OK. When engineers reviewed the raw request logs, they...
-
Silent Omissions: Why Ambient AI Scribes Drop Clinical Negatives
Your health system deployed an ambient AI scribe across outpatient clinics to capture patient-doctor dialogue and generate clinical SOAP notes in real time. Clinicians love the speed. Note turnaround times drop significantly. Then a clinical documentation audit uncovers a systemic issue: during...
-
Phantom Policies: Why AI Prior-Auth Agents Suffer Vector Drift
Your health system automated prior authorization using an autonomous agent to read charts, query payer policy databases, and assemble submission packets. The system runs smoothly. The API calls succeed. Then your denial rate spikes by 34%. Claims examiners discover that your AI agent cited a...
-
The 50mg Bug Why Autonomous AI Agents Break Clinical EHR Data
Your engineering team deployed an autonomous clinical agent to transcribe patient check-ins, extract medical history, and push structured records directly into your Electronic Health Record via standard FHIR APIs. The pipeline executes smoothly. The API gateway returns HTTP 200 OK across…
more -
Silent Docket Drift: Why AI Agents Miss Statutory Court Deadlines
Your firm deployed an autonomous litigation assistant to track court dockets, extract trial dates, and push statutory filing deadlines into your calendar. The sync pipeline runs daily. The status logs read green. Then the court issues a default judgment against your client. Outside counsel...
-
The Privilege Leak: Why Vector DBs Break Attorney-Client Confidentiality
You connect an enterprise LLM to your internal knowledge base to speed up employee workflows. Two weeks later, a non-legal employee asks about product liability risks, and the bot summarizes an internal memo written by outside counsel detailing an ongoing regulatory investigation. The...
-
The $14M Due Diligence Blunder: Why Basic RAG Fails in Legal Ops
Imagine running automated due diligence across thousands of vendor agreements or a 500-page acquisition filing. Your AI review agent runs, parses every document, and presents a clean report with zero red flags. Two months post-close, litigation hits. You discover the agreement contained a $14...
-
Why AI Agents Silently Fail in Production (Prompt & Schema Drift)
Your deployment pipeline is green. Your monitoring dashboards in Datadog report 100% uptime. Your API gateway latency is sitting comfortably at 180ms.Yet, over the last 30 days, your enterprise multi-agent workflow’s end-to-end task completion rate quietly deteriorated from 92% to 71%.No one...
-
"Dumb RAG" & Context Flooding: Why Big Context Windows Fail
You dumped all your company docs into a vector database, and now your AI agent is quoting 2022 pricing to your enterprise clients. In Episode 34, Maya breaks down the "Dumb RAG" and Context Flooding trap: why treating an LLM context window like a hard drive instead of RAM causes model thrashing,...
-
Silent Tool Call Failures: How AI Agents Corrupt Production Data
Your AI agent didn't throw a 500 error. It got an empty response from an API, hallucinated a workaround, and corrupted 50 customer records without alerting anyone. In Episode 33, Maya breaks down the silent tool call failure in multi-agent architectures: why standard APM and SRE tools fail to...
-
Why Specialty Clinics Lose $150k to Intake Hold Times
Specialty medical clinics lose up to 34% of inbound calls to 15-minute hold times—costing practice groups over $150,000 annually in lost patient volume. In Episode 32, I break down the administrative bottleneck in healthcare intake: why front desk staff are forced to act as human middleware…
more -
Solving the "Confused Deputy" Problem in AI Agent Architecture
Your AI agent wasn't breached by a hacker. It was tricked into executing an admin API call because you gave it your developer permissions. In this episode, I break down the "Confused Deputy" vulnerability in autonomous AI agents: why system prompts fail as security boundaries, how non-human...
-
Why AI Coding Agents Leak Secrets in CI/CD (And How to Stop It)
Your developer didn't leak an API secret to GitHub. Their autonomous AI coding agent generated a mock key, filled in a real one, and opened a Pull Request. This episode breaks down the growing security risk of AI coding agents in CI/CD pipelines: why agents hardcode real credentials into test...
-
Why 90%+ AI Benchmark Scores Fail in Production
Your model scored 92% on public benchmarks, but it’s failing on 30% of your real-world production tickets. This episode breaks down the benchmark saturation paradox: why static academic benchmarks fail to predict production accuracy, how clean training data creates false confidence, and how to...
-
Why Your RAG System Serves Wrong Answers with 0.94 Similarity
Your RAG system didn't hallucinate. It retrieved an outdated document with a 0.94 similarity score and answered with total confidence. This episode breaks down the RAG Freshness Trap: why high vector similarity scores frequently mask stale context, how embedding drift happens quietly in…
more -
The One AI Question Every Hospital Board Should Ask in 2026
Hospital boards approve AI slides with a smile and almost never ask the question that reveals whether any of it is safe. Show me one real case from last month where AI touched a patient or money, and walk me through the logs. If the room goes quiet, that is the gap. This episode covers what a…
more -
"We're Just Piloting AI Agents." It's Touching Real Patients. That's Production.
Hospitals love saying they are just piloting AI agents, while those agents are already touching real patients and real money. Internally it is only a test. Externally, real people are getting real outcomes from a system nobody officially owns. This episode breaks down why pilot is a permission…
more -
Should Doctors Trust AI Scribes With Their Notes? Read This Before You Sign.
AI scribes are saving doctors hours every week, and they can quietly rewrite what happened in the room. A patient says they had chest pain. The AI writes, denies chest pain. You sign, and now your own note contradicts the visit. This episode breaks down why scribe errors are more dangerous than...
-
Every "AI in Healthcare 2026" Talk Says the Same 3 Things. All Incomplete.
Every AI in healthcare 2026 talk says the same three things. Ambient scribes fix burnout, diagnostic AI catches what humans miss, predictive models save lives. All true, and all incomplete. The reliable tools are boring. The worries live one layer up, in shadow AI, model drift, and architectures...
-
81% of Doctors Use AI. Almost None Can Say Where It Lives in Their Clinic.
81% of doctors now say they use AI in their practice, and almost nobody can list where the AI is actually hiding in their own clinic. Ambient scribes, decision support, triage tools, patient chatbots. The pitch is less documentation and less burnout. The reality is that you are now responsible…
more -
AI Agents Are Live in Hospitals. Nobody Knows What They Did Yesterday.
AI agents are finally live in hospitals in 2026, and most teams have no idea what those agents actually did yesterday. One hospital wires an agent into the call center. It listens, updates the chart, and decides whether to book a visit. For a week it is magic. Then a regulator asks why the agent...
-
Reddit, YouTube, LinkedIn All Lie About AI Agents. Read All Three
Reddit thinks AI agents are chaos. YouTube thinks they are a solved pattern. LinkedIn thinks they are already transforming everything, and nobody there wants to show you the incident reports. If you ship agents into clinics, banks, or firms, you cannot afford to live in just one of those worlds....
-
Your AI Agents Watch 13 Platforms. None of Them Know What You Ship.
You gave your AI agents access to thirteen platforms and now your strategy meeting is everyone drowning in insights nobody can use. One agent scrapes what is trending, one clusters themes, one writes a beautiful morning deck. It feels powerful, right up until you realize the whole stack has no…
more -
Epic Turned On AI Across Your Hospital. Who Owns It When One Feature Misbehaves?
Epic switched on a wave of AI across clinical care, the patient portal, and the revenue cycle. Art drafts notes and queues orders. Emmie answers patients in MyChart. Penny touches claims and appeals. On a good day, all of it saves time. This episode asks the question that actually decides whether...
-
Three Calls Your AI Agent Can’t Afford to Mess Up
Three phone calls. Three industries. Same AI agent failure pattern. In this episode, Maya walks through: Clinic – a caller reports chest pain, and the AI receptionist treats it like a routine scheduling request. Missing escalation rules turn clinical risk into a calendar slot. Bank – an AI agent...
-
Reddit vs Reality: 3 AI Agent Failure Modes
Reddit keeps screaming that “AI agents are broken.” In this episode, Maya walks through three failure modes that show up again and again when teams ship agents into real workflows: The retry storm – everyone adds retries to “make it reliable”, and quietly creates a self‑inflicted DDoS and a…
more -
You Don’t “Use” AI Agents. You Deploy Them.
Most teams think they “just use AI tools.” Regulators don’t care who trained the model. They care that your AI touches patients, money, or rights. If you plug an AI receptionist into a clinic, let an AI agent nudge credit decisions, or route privileged calls through an AI assistant at a law firm,...
-
You Were Sold an AI Agent. You Got a Chatbot.
You paid for an AI agent. You got a chatbot in a trench coat. A chatbot answers. An agent acts. The real difference is whether it can finish work, hold its place across steps, and recover when one of them fails. This episode breaks down the three step test that separates a real agent from a…
more -
Your Company's Private Data Is Already Inside a Public AI Model
Right now, someone on your team is pasting source code, customer records, or an internal document into a public AI model to move faster. It works. It also just left your company. Every prompt is a data export, and nobody logged it. This episode breaks down why the real gap is the missing…
more -
An AI Agent Deleted an Entire Company's Database in 9 Seconds
An AI agent was given access to a production database. It hit one thing it did not understand, guessed instead of stopping, and deleted the database along with every backup. Then it apologized. This episode breaks down why the agent destroyed everything instead of stopping, the three controls…
more
We're indexing this podcast's transcripts for the first time — this can take a minute or two. We'll show results as soon as they're ready.
No matches for "" in this podcast's transcripts.
No topics indexed yet for this podcast.
Loading reviews...
ABOUT THIS SHOW
Enterprise AI infrastructure explained by someone who has seen what breaks in production. Every episode breaks down one concept that matters when you are running AI at scale in regulated industries. LLMOps. Orchestration. Governance. Compliance. Observability. The stuff between the models and the business logic that nobody talks about until something goes wrong. New episodes three times a week.
HOSTED BY
Maya Chen
CATEGORIES
Loading similar podcasts...
Frequently Asked Questions
How many episodes does Maya Builds AI have?
Maya Builds AI currently has 61 episodes available on PodParley. New episodes are automatically indexed when they're published to the podcast feed.
What is Maya Builds AI about?
Enterprise AI infrastructure explained by someone who has seen what breaks in production. Every episode breaks down one concept that matters when you are running AI at scale in regulated industries. LLMOps. Orchestration. Governance. Compliance. Observability. The stuff between the models and the...
How often does Maya Builds AI release new episodes?
Maya Builds AI has 61 episodes. Check the episode list to see recent publication dates and frequency.
Where can I listen to Maya Builds AI?
You can listen to Maya Builds AI on PodParley by clicking any episode. We provide an embedded audio player for direct listening, and you can also subscribe via your preferred podcast app using the RSS feed.
Who hosts Maya Builds AI?
Maya Builds AI is created and hosted by Maya Chen.
URL copied to clipboard!