Frontier AI for clinical intelligence.

We are building the substrate for clinical evidence — turning unstructured hospital data into a standardised, queryable format, and the superintelligence reasoning layer on top of it.

Read the whitepaper Learn more →

Premise

Hospitals are the most data-rich institutions on earth — and the least data-fluent.

Modern medicine generates an extraordinary signal: every visit, every scan, every note. Yet almost none of that signal is usable. Data lives in PDFs, dictated narratives, and inconsistent EHR fields - readable to clinicians but opaque to machines.

Hospitals know more, in aggregate, than any other organisation - but answer less. Population-scale questions remain unanswered. Operational decisions are made on lagging dashboards. Research is gated by months of manual abstraction.

Amorphous AI exists to close that gap. We are building intelligent systems that transform unstructured data into clinical insights.

50 PB

generated per hospital, per year

80%+

of it is unstructured text

97%

of clinical data goes unused

Pipeline

From unstructured notes to auditable answers.

Follow a single discharge note through the pipeline — raw data is extracted, standardised against clinical ontologies, queried by agents, and returned as a reproducible result. Every step keeps its own provenance trail.

Latency: ~2 m
Entity Recall: 97.8%
Ontologies: SNOMED · LOINC · ICD-10 · RxNorm
Audit Artifacts: SQL, Python Code, Visualisations

01Input

Discharge Note

Patient denies chest pain.
Hx HTN, DM2 — on metformin 500mg BID.
Troponin 0.8 ng/mL, elevated.

02Structure

Data Engine

HypertensionSNOMED CT 38341003

Type 2 diabetes mellitusICD-10 E11

Metformin 500mgRxNorm 6809

Troponin (elevated)LOINC 6598-7

03Reason

Striata

Q: Compare HbA1c trajectory — insulin vs oral antihyperglycemics.

↳PlanningCohort Definition

↳Data QuerySQL Code

↳AnalysisWelch's t-test

↳VisualisationBar Chart

↳ReportMethodology & Findings

04Output

Auditable Report

Δ HbA1c · 12 months

Insulin

−0.84%

n = 218

Oral

−0.18%

n = 194

Δ = 0.66 ppp = 0.005

Striata

Ask a question. Watch the work.

Striata resolves umbrella terms, writes the SQL, generates the Python, picks the statistical test, and produces a written finding. Every step is inspectable.

Striata

QuestionsProcess Document

‹Metformin Prescription Rates by Age

ChatSuccess

Fibrocalculous pancreatic diabetesSNOMED CTDrug-induced diabetes mellitusSNOMED CTSecondary diabetes mellitusSNOMED CTMetformin oral productRXNORMmetformin hydrochloride 500 MG Oral TabletRXNORMmetformin hydrochloride 1000 MG Oral TabletRXNORMShow all 160 codes

⌕

Dataset Query

⌄

▦

Visualization

⌄

◱

Report Draft

Metformin Prescription Rates by Age

⌄

I've completed the analysis and prepared a report titled "Metformin Prescription Rates by Age". You can view the full findings in the Findings panel.

⌄ Show Suggestions

⌘ Ask a follow-up…

↑

Findings

Create ReportExplore Data

Metformin Prescription Rates by Age

Summary

Of 18,402 diabetic patients, metformin use peaks in the 45–64 bracket at 78.1%, then drops to 58.3% (65–79) and 31.2% (80+) — an absolute decline of 46.9 pp from peak, consistent with deprescribing for renal impairment.

Figure 1

Metformin Prescription Rate by Age Group

80%

60%

40%

20%

71.4%

<45

78.1%

45–64

58.3%

65–79

31.2%

80+

Age Bracket

<45

45–64

65–79

80+

Sample Questions

The questions your team has been carrying for years.

Data & Analytics Officer asks Striata:

"How much of our analyst backlog could we clear if every question returned in minutes, with auditable SQL and code?"

Lab Notes

Making sense of unstructured data in hospitals

We don't ship improvements unless we can quantify them.

Inside the whitepaper: how the Data Engine standardises clinical records, the multi-agent architecture behind Striata, and the benchmark infrastructure and results that drive development.

Read Online ↗

Backing

$2M pre-seed, backed by operators.

Led by Specialist VC and Calm/Storm, with backing from Z Fellows, BADideas.fund, Outlast Fund and Plug and Play, alongside leading US and EU angels.

Investors

Specialist VC Calm/Storm Z Fellows BADideas.fund Outlast Fund Plug and Play

Contact

Modernizing your hospital's data infrastructure?Let's talk.

For partnerships

hello@amorphous.health

Get in touch