Burna AI · Launch · May, 2026

We're rebuilding the safety architecture under every cancer drug approval.

Twelve specialized agents. Citation on every grade. Two patents filed.

Nnenna JohnFounder and CEO, Burna AIMay, 2026

burna.aiMember, CancerX (HHS Cancer Moonshot)

TLDR

Adverse event grading is how cancer trial safety data is built, one event at a time. The workflow has not changed in 20 years. Burna AI is the safety and data quality platform that does change it. Twelve specialized agents in a patented cascading constraint pipeline. Every grade carries a citation to source clinical text and a CTCAE criterion. Two patents filed. Accepted into a top-tier cancer center accelerator program (June 2026 cohort, 1,200+ patient charts across three campuses). Atlantic Health System engagement underway. Member of CancerX. HIPAA compliant, SOC 2 certified, 21 CFR Part 11 aligned. Eligible for the FDA CDER Emerging Drug Safety Technology Program.

About us

A founder's letter, signed in the open.

I'm Nnenna John, Founder and CEO of Burna AI. I spent 19+ years in platform engineering, including six years at Airbnb building core infrastructure and taxation systems that served 150M+ users across 220+ countries. Before that: JP Morgan Chase, building regulatory-compliant systems that processed billions in transactions. Roku. Expedia.

I left to build Burna AI because of a question I could not stop asking. Cancer drug approvals are the highest-stakes decisions in regulated medicine. They rest on safety data. That data is graded one adverse event at a time, on a workflow built for a paper-and-fax era. Why does the architecture under it look the way it does?

My co-founder Paschal Ezeugwu joined as Lead Engineer. He brings deep EHR expertise. We built Burna AI together, starting July 2025. I personally architected the grading pipeline. The platform serves cancer centers and pharma sponsors. Today we are launching publicly.

The problem

A workflow that has not changed in twenty years.

Manual CTCAE grading takes 15 to 20 minutes per adverse event. There are 850 criteria in CTCAE v6.0, across 26+ organ system classes, with quarterly regulatory updates. Coordinators and clinicians cross-reference those criteria against patient notes by hand, one event at a time. The workflow runs on PDF lookups, paper forms, fax, and email.

The peer-reviewed evidence on what this produces is plain. Inter-rater agreement on attribution: kappa 0.59 to 0.68 (Hong et al., 2020). Attribution change between investigators and central review: 31 to 36% (Hillman et al., JCO 2010). The 2019 FDA-NCI workshop characterized the state of the workflow in its own terms.

In multi-site trials, query cycles between sponsors, CROs, and investigators stretch for weeks. Grading inconsistencies surface at the next monitoring visit, not at the point of care. By the time a sponsor sees a Grade 3+ event, the patient is in the next cycle.

At Memorial Sloan Kettering's scale (1,800+ studies, 400+ coordinators, 800+ PIs), the labor chain around adverse event grading and reporting runs $40 to $60M annually. Across the 70+ NCI-designated cancer centers, the aggregate is multiples of that.

Cancer drug approval rests on this data.

0.59→ 0.68

Inter-rater kappa

Hong et al., 2020.

31–36%

Changed by central review

Hillman et al., JCO 2010.

$40–60M

Annual labor chain · MSK scale

1,800+ studies, 400+ coordinators.

“Sub-optimal,
unreliable,
inefficient.”

2019 FDA-NCI workshop

Verdict on the current state.

What we built

Twelve agents. One constraint pipeline.

The engine. The real-time architecture. The two products that ride it.

The engine

Twelve specialized agents in a patented cascading constraint pipeline.

Burna AI's grading engine is twelve specialized agents operating in a patented cascading constraint pipeline. The pipeline spans extraction, verification, resolution, standardization, taxonomic matching, differential analysis, temporal tracking, documentation integrity, evidence synthesis, causal attribution, confidence calibration, and regulatory encoding. Each agent's output bounds the valid output space of every agent downstream. The grade at the end of the chain is the only grade consistent with every prior agent's findings.

The output space is constrained to valid CTCAE grades. Citation to source clinical text and CTCAE criterion is mandatory: the engine cannot produce a grade without it. Attribution runs on WHO-UMC and Kramer algorithms with probability scores per drug (e.g., "Oxaliplatin: 72% probability"), across 42 pre-built oncology regimen profiles. The math is visible. A clinician makes every final decision.

This is reasoning under constraint, not classification. Two patents filed.

Real-time from the ground up

Every grading decision propagates instantly.

Every grading decision propagates instantly. Grade 3+ events trigger immediate notifications. In multi-site trials, grading inconsistencies surface in real-time across all sites, not at the next monitoring visit. Most EDC and pharmacovigilance platforms were designed for batch capture in an era when trial data was reviewed weekly. Burna AI was built real-time from day one.

Two products, one engine

CTCAE AI for cancer centers. The Protocol Safe for sponsors.

Product 1

CTCAE AI

Serves NCI-designated cancer centers and academic medical centers. Three input modes: ambient capture of live encounters, manual paste of existing notes, and batch retrospective review for chart audits, real-world evidence studies, and pharma safety surveillance. SMART on FHIR integration for Epic, Oracle Health, and Veeva Vault EDC.

Product 2

The Protocol Safe

Serves pharma sponsors. An isolated agent deployed inside the sponsor's own cloud infrastructure ingests the trial protocol. When a site submits a clinical note for attribution grading, the Burna AI engine queries that isolated agent for the trial drug's expected adverse event profile. The engine receives back only the attribution signal. The protocol never leaves the sponsor's infrastructure. Content Boundary Middleware validates every outbound response. Fail-closed by design.

The white space

Upstream of every existing pharmacovigilance platform.

Every existing pharmacovigilance platform (Oracle Argus Safety, ArisGlobal LifeSphere, Veeva Vault Safety, and the Vault Safety AI Agents launched April 2026) operates downstream of the attribution decision. They manage the ICSR lifecycle after attribution has already been made.

Nobody helps the investigator make the attribution decision at the point of clinical encounter. That is where Burna AI operates. That is the white space, and it is large.

Why now

The Flatiron parallel, AI-native and two-sided.

The most regulated, highest-stakes data in cancer drug development, with no dominant incumbent.

Flatiron Health built the data quality layer for oncology real-world evidence and sold to Roche for $1.9 billion. Pre-AI. Single-sided. Nobody has built the equivalent for oncology safety data, the most regulated, highest-stakes data in cancer drug development. Burna AI is the AI-native, two-sided version. Provider SaaS for cancer centers. Pharma data quality services, Protocol Safe, and postmarket pharmacovigilance for sponsors.

The regulatory tailwind is real. FDA's CDER Emerging Drug Safety Technology Program creates a defined pathway for AI safety infrastructure. CTCAE v6.0 added 800+ criteria changes. Pharmacovigilance processing demand is growing 12 to 15% CAGR on a base of $8 to 10B+ globally.

$15 to 20B+

Total addressable market

Across the drug lifecycle. No dominant incumbent.

12 to 15%

Pharmacovigilance CAGR

On a base of $8 to 10B+ globally.

800+

CTCAE v6.0 criteria changes

Mapped ahead of industry adoption.

Traction

Ten things we can prove today.

Each line is a verifiable, dated commitment. Conservative on external claims, specific on what is signed and shipping.

Patents
Two patents filed on the agentic grading pipeline.
Accelerator
A top-tier cancer center accelerator program: accepted into the June 2026 cohort. 1,200+ patient charts across three campuses.
Atlantic Health System
Technology Subcommittee demo completed April 9. No-cost pilot on the table. Cardiology interest surfaced unprompted, confirming the platform applies beyond oncology.
Memorial Sloan Kettering
Direct engagement confirming the $40 to $60M annually labor chain at MSK scale.
CTCAE v6.0
Mapped ahead of industry adoption.
CancerX member
HHS, ONC, Moffitt Cancer Center, Digital Medicine Society.
Compliance
HIPAA compliant, SOC 2 certified, 21 CFR Part 11 aligned.
FDA EDSTP
Eligible for the FDA CDER Emerging Drug Safety Technology Program. Application planned for June to July 2026.
Advisors
20+ advisors across clinical, academic, pharmaceutical, federal health IT, and commercial domains.
Internal testing
Ongoing internal testing shows strong agreement with expert clinicians on edge-case adverse events.

The advisory board

Twenty-plus people who hold us to the work.

Across clinical practice, pharmaceutical development, federal health IT, academic informatics, and commercial strategy.

Clinical

Aman Opneja, Mary Morison Saltz, Usman Shah, Manuj Agarwal.

Pharmaceutical development

Stefan Gluck, Andrea Pirzkall, Michel Azoulay, Kira Sentrenko, Jide Sotunbo.

Federal health IT

Milad Bahrami.

Academic informatics

Mark Schweitzer.

Commercial strategy

Rebecca Greene, Azita Hamedani, Laurence Girard, Oliver Overheu, Ryan Joswick.

Dr. Stefan Gluck presents Burna AI at ASCO in late May.

Asks

Three conversations we are open to today.

Direct lines for cancer center and pharma leaders, conference attendees in May, and investors in oncology drug safety infrastructure.

Ask 1

Cancer center research operations and pharma pharmacovigilance leaders.

If you lead research operations at an NCI-designated cancer center or AMC, or pharmacovigilance at an oncology pharma sponsor, the design partnership conversation is open. burna.ai or nnenna@burna.ai.

Email nnenna@burna.ai

Ask 2

Conference connections in May.

Burna AI is on the ground at SCOPE X (May 18 to 19, Boston), Bio-IT World (May 20 to 21, Boston), and ASCO (May 29 to June 2, Chicago). Dr. Stefan Gluck presents in Chicago. Book time directly: https://calendly.com/d/cwh9-yh2-z77/burna-ai-introduction.

Book time on Calendly

Ask 3

Investors in oncology drug safety infrastructure.

One more thing: we are currently fundraising and bringing on a small group of investors before this stage closes in June. Happy to share details if it is interesting to you.

Reach out to Nnenna

Contact

nnenna@burna.ai

Burna AI, Inc. · Delaware C-Corp · Atlanta, GA

Member, CancerX (HHS Cancer Moonshot Initiative)