Matt Margolis, MS
Founder and CEO
Researcher from the diagnostics industry and a named inventor on an IGF1R mutation patent. Matt built Silogram end to end.
About Silogram
The record between the analysis and the evidence payers read.
Silogram links the code that produces a study's numbers to every document that reports them, from the first abstract to the payer dossier. We start with mid-market pharma, biotech and diagnostics teams, who carry big-pharma evidence loads with small teams.
Our mission
Get trustworthy evidence to the people who decide on coverage, sooner.
Medicines and diagnostic tests reach patients once payers, guideline panels and HTA bodies have the published evidence they ask for. Between a locked analysis and that evidence sits a lot of internal work: copying numbers into documents, checking them, re-checking them after every rerun, and gathering sign-off.
Silogram takes the copying and re-checking off people's desks and keeps the record as it goes: numbers linked to the code that produced them, every change and signature on the audit trail, and a forecast of when each document will be ready for the decision it supports.
What we believe.
A number should lead back to its code
Every value in a paper, a CSR or a dossier should be traceable to the analysis that produced it, and should change only when that analysis does.
People decide; agents suggest
Models and agents draft and check. Every edit they make waits for a person to accept it, under that person's name, on the record.
Your data stays yours
We read code and aggregate outputs. Your documents never train AI models, and code that touches identifiable patient data runs only inside your own network.
Say what is documented
We keep what the evidence shows apart from what we hope. Our pages cite their sources, and our comparisons are dated and open to correction.
Why now.
More scrutiny on every number, and agents writing more of them.
Numbers drift between documents
In one study of phase 3 and 4 trials, 20% reported a different primary outcome value in the journal than on ClinicalTrials.gov.1
AI use must be disclosed
ICMJE's January 2026 recommendations give AI its own section: authors disclose AI use in the cover letter and the manuscript, and stay responsible for the output.2
HTA asks for the code
EU joint clinical assessments, required for new cancer medicines since January 2025, ask for CSRs, protocols and SAPs, and for program code and output where an analysis isn't a standard method.3
Unpublished numbers go public
NICE and CADTH no longer routinely redact clinical data awaiting publication, so the numbers in an HTA submission have to match the paper that follows.4
Founder-built, from the diagnostics industry.
We're hiring
Build Silogram with us
The first roles are a founding engineer and a publications and regulatory writing lead.
See open rolesMore about Silogram
How we handle security
Audit trail, e-signatures, approval locks, the PHI guard and sandboxed reproduction, with the status of our assurance work stated plainly.
Security and complianceHow we compare
A dated, sourced comparison with publication-planning suites, AI writers and code-to-document tools, including what Silogram doesn't do.
CompareTalk to us
Pilots, security questionnaires, press and partnerships.
ContactSources
- 1Hartung et al. Ann Intern Med. 2014. PubMed 24687070.
- 2ICMJE. Recommendations for the conduct, reporting, editing and publication of scholarly work in medical journals (January 2026).
- 3HTA Coordination Group. Guidance on the JCA dossier template for medicinal products (2024).
- 4CADTH, ICER and NICE. Joint position statement on confidential data (April 2023).