One-Page Overview

How Nytel Works

AI agents that automate the coordination work in clinical trials — sitting above existing systems, not replacing them.

The Problem

Clinical trials run on coordination across CTMS, eTMF, EDC, email, Slack, and spreadsheets. During active conduct, formal systems like eTMF and EDC are often populated after the fact for regulatory purposes — the real source of truth lives in email and people's heads. 70% of trials are delayed, and enrollment delays cost $500K+ per day.

Existing platforms (Medidata, Veeva, IQVIA) optimize within their own silos. The work between systems — chasing documents from sites, reconciling data, tracking who owes what — remains manual.

Architecture

Nytel is designed as an orchestration layer above existing clinical systems. Specialized AI agents each handle one operational workflow, sharing a common knowledge layer. Every agent follows the same pattern: Retrieve data → Analyze with deterministic rules → Create auditable outputs.

HUMAN SUPERVISION Clinical Ops Lead Study Manager Medical Monitor NYTEL ORCHESTRATION LAYER Document QC Classify, verify, flag ALCOA++ rules Site Activation Track CDA → go-live Milestone tracking Follow-Up Escalate, batch, pause Smart cadence Monitoring Visits, enrollment, data Anomaly detection Safety + more agents EXISTING CLINICAL SYSTEMS eTMF CTMS EDC Email Slack Spreadsheets

Key Design Principles

Starts where the data is

Email, Slack, spreadsheets — not API integrations. No IT procurement needed to begin. Direct system APIs come in Phase 2.

Deterministic compliance

Quality checks use rules and code, not generative models. Every decision is auditable. LLMs handle document and communication understanding only.

Human-in-the-loop

Agents execute; humans supervise. Low-confidence OCR triggers review. Safety exceptions bypass automation. Teams can pause and override.

Entry Point: Site Activation

The first workflow targets site activation — CDA execution to site go-live. This is where timelines slip first, where document-chasing is most acute, and where a measurable KPI (CDA-to-activated time) makes success easy to demonstrate. A small set of document types — investigator CVs, medical licenses, GCP certificates, FDA 1572 forms, delegation logs — drives the majority of QC work during activation.

From there: TMF readiness → site monitoring → issue triage and CAPA. Each workflow added reuses the same knowledge layer, so marginal effort drops as the system matures. Institutional knowledge — which sites are slow, which sponsors have unusual requirements — accumulates as data rather than disappearing with every staff change.

Want the full picture?

Interested in working together?

Nytel is working with clinical operations teams to pilot AI-driven trial workflows. If you're dealing with site activation delays, document QC bottlenecks, or inspection readiness gaps — let's talk.