Four of these have papers less than a year old. Three have zero direct competitors. One is just waiting for someone to build the factory.

What just became possible

TB drug susceptibility testing takes 2 to 4 weeks today, because you have to grow the bacteria and watch what kills them. Solvatochromic trehalose probes get metabolized by live mycobacteria and light up in days. Same phenotypic readout, no genetic guesswork, works against drugs that don’t exist yet.

Why now

The Tre-DST paper hit bioRxiv on June 23, 2025. It showed the probes distinguish drug-susceptible from drug-resistant TB in three doubling times. Before that, every fast TB test was molecular, which means it only flags resistance patterns we already know about. Phenotypic was the gold standard and slow. Now it can be both.

What you’d build

A cartridge-based phenotypic DST kit in the GeneXpert form factor. Sells to national TB reference labs in India, South Africa, and Brazil at $50 to $100 per test, replacing $100+ liquid culture or whole genome sequencing. Target market: $500M to $1B by 2030.

Who’s already moving

The Tre-DST and TRACeR-TB academic groups have the chemistry. Skysong Innovations has an early label-free platform. HaystackAnalytics does genome sequencing, not phenotypic. We searched. No startup has the cartridge format yet. Cepheid dominates molecular at 50% market share but doesn’t compete on second-line DST.

The gap

Probe stability in a shelf-stable cartridge that survives 30°C+ field conditions for six months. Fluorescence readers cheap enough for resource-limited labs. WHO prequalification. $167M in TB diagnostic grants flowed in 2023, and the WHO target product profile for next-gen DST was published in August 2024. The runway is open.

What just became possible

General-purpose LLMs hallucinate 17 to 33% of the time on legal citations, by independent benchmark. That kills them for compliance, where wrong answers carry liability. Combining LLM recall with a symbolic logic layer (Logic-LM, SatLM) gets you constrained decoding and statistical guarantees. Recent benchmarks hit 92.6% precision on financial compliance tasks.

Why now

The survey paper on RAG, reasoning, and agentic hallucination mitigation dropped on arXiv on November 4, 2025. It mapped the path to provable error bounds. EU AI Act enforcement starts in 2026. FATF guidelines for AI in AML are being rewritten. Tier-1 banks need something they can sign off on legally.

What you’d build

An enterprise API for entity and relationship extraction with a contractual sub-1% error rate. Pricing: per-document, six figures annual contract. US banks alone spend $25B/year on AML compliance and ate $4.5B in fines in 2024.

Who’s already moving

SAIFR, 4CRisk.ai, SymphonyAI Sensa do AI-native RegTech but none offer provable error bounds. Quantexa and ComplyAdvantage are probabilistic. LexisNexis and Thomson Reuters have the brand but the published hallucination rates. We checked the white space. Nobody is selling statistical guarantees on extraction quality.

The gap

Scaling symbolic reasoning to enterprise document volumes. Building domain-specific knowledge graphs for FIBO and ISO 20022. Getting one Tier-1 bank to sign a pilot through the HKMA or FCA sandbox, and a Big Four firm to validate the error rate. The talent costs $165k to $230k for neuro-symbolic specialists and the ratio is 8:1 open roles to qualified candidates.

What just became possible

eGenesis ran a 69-gene-edited pig kidney that lasted 271 days in a human. United Therapeutics has FDA IND clearance for the first xenokidney clinical trial. The science of immune rejection and coagulation in 10-gene-edited pigs is solved enough to start Phase 1/2. The factory to make these organs at scale is not.

Why now

The bioRxiv paper on 10-gene-edited specific pathogen-free pigs landed on May 10, 2025. It confirmed survival in non-human primates with mitigated immune rejection and coagulation issues. United Therapeutics is building a 125-organ-per-year facility. The first commercial xenotransplant will need a contract manufacturer before 2027.

What you’d build

A GMP-compliant bioprocessing platform for SPF, gene-edited pig organs. Closed-system, sterile, reproducible. Customers are United Therapeutics, eGenesis, Revivicor, Qihan Biotech, Sino-Clone. They’re the buyers, not the competition. The global xenotransplantation market is $17.13B in 2026 heading to $23.51B by 2030, with 42% from pig organs.

Who’s already moving

Samsung Biologics and Bora Biologics do GMP cell therapy, not whole organs. United Therapeutics is building its own facility, which tells you exactly where the white space is. Nobody runs xenotransplant manufacturing as a service. Every clinical-stage xeno company breeds and procures organs in-house with non-standardized veterinary teams.

The gap

FDA wants 50-year sample storage on every donor animal. Veterinary surgical teams that meet GMP. Closed-system bioreactors sized for whole organs, not cell suspensions. NIH funded U19AI191396, R01AI144522, and S10RR027050 in 2024 to 2026 for this. Whoever builds the factory wins by default.

What just became possible

Most rapid malaria tests detect HRP2, a P. falciparum antigen. The parasite has been deleting the hrp2 and hrp3 genes. In Eritrea and Djibouti, 60 to 83% of strains carry the deletion. LDH-based rapid tests detect both falciparum and vivax, including hrp2/3-deleted strains. The latest field evaluation caught 100% of deleted-strain cases.

Why now

The medRxiv evaluation published on July 3, 2025 made the data unambiguous. WHO updated its guidance: switch from HRP2 to LDH when more than 5% of cases are missed. Djibouti switched in 2021. Ethiopia and Eritrea are mid-transition. 300 million RDTs are procured annually in sub-Saharan Africa and a large share are now the wrong test.

What you’d build

A field-deployable LDH-based RDT with WHO prequalification, thermal stability at 30°C+ for six months, and unit cost of $0.50 to $1.00. Buyer is the national malaria control program in each endemic country. Procurement runs through UNICEF and the Global Fund.

Who’s already moving

Rapigen’s BIOCREDIT Pf/Pv is in WHO prequalification and already deployed in Djibouti. Abbott Bioline is the HRP2 incumbent shifting to combo. Premier Medical sells First Response in India. AccessBio exited the malaria RDT market in 2020. The field is open for one or two specialists who can hit price parity on LDH.

The gap

LDH enzyme stabilization at tropical temperatures is the hard biochemistry. WHO prequalification takes 12 to 18 months minimum. National tender requirements often favor incumbents. The market is $950M in 2026 heading to $1.45B by 2035, with Asia Pacific growing 7.8% annually. Unitaid and TDR are funding the transition right now.

What just became possible

Pyrethroid-only bed nets stopped working in 78 of 88 malaria-endemic countries because mosquitoes evolved. Dual-active-ingredient nets (chlorfenapyr or pyriproxyfen plus pyrethroid) cut malaria by 20 to 50% in resistant areas. Interceptor G2 and Royal Guard are both WHO prequalified and shipping at scale.

Why now

The Research Square trial published on May 10, 2024 confirmed field efficacy against pyrethroid-resistant Anopheles gambiae. The Global Fund and Unitaid funded $66M to deploy 56 million dual-AI nets across 17 countries. Market share jumped from 59% to 80% in one year. The 2024 to 2026 grant cycle locked in the transition.

What you’d build

The net itself is not the play. BASF makes Interceptor G2. Disease Control Technologies makes Royal Guard. Sumitomo and Vestergaard have variants. The play is downstream: resistance surveillance, durability monitoring, and AI-enabled vector tracking so procurement can route to the right product before the next resistance wave hits. Johns Hopkins CBID is hiring for exactly this.

Who’s already moving

IVCC supports product development. Target Malaria works on genetic vector control. Several academic groups run resistance surveillance. No commercial company sells procurement-grade resistance maps to NMCPs or Global Fund procurement. The buyers spend $1.5B to $2.7B per year on nets alone.

The gap

Chlorfenapyr resistance will emerge. The history of pyrethroids says three to five years. The surveillance infrastructure to detect it early and route procurement accordingly does not exist as a commercial product. That is the gap.

See you next week.