Clinical trial monitor
Here's what AI can do for this role — and what still needs a human. Built straight from ENSEK's own job advert, running live on my_db.fisher_demo.clinical_trial_visits. Not a slide about AI. The job, getting done.
Every line on the left is lifted from ENSEK's actual job ad. If a card lacks a harvested JD line, it is omitted. On the right is the AI doing it — with eligible cards running live against the warehouse and offline inspection clearly labelled in the workspace.
“site enrolment vs target — completion velocity by site and study, used to trigger enhanced monitoring or site escalation.”
How is site enrolment progressing across studies — which sites are tracking to target and which are lagging the completion velocity?
bar chart“protocol deviation rate by site and category — the GCP monitoring heat map the CRA uses to target corrective action visits.”
Where is the protocol deviation rate highest by site and category — the CRA monitoring heat map for targeted follow-up?
table“outstanding data queries by site — source data verification gap before database lock. ALCOA+ compliance signal.”
Which sites have the highest outstanding data query counts — the source data verification gap that needs CRA follow-up before database lock?
table“visit completion rate by visit type — dropout and missed-visit pattern across the trial schedule. GCP protocol compliance signal.”
How does visit completion rate vary by visit type — where do dropout and missed-visit patterns cluster across the trial schedule?
bar chart“critical deviation trend by month — safety signal watch for CRA and sponsor escalation under ICH E6 (R2) mandatory reporting.”
How is the critical deviation rate trending by month — are safety signals improving or escalating across the monitoring period?
tableThe honest other half. AI does the analysis; a person owns the decision — especially where regulation, fairness and accountability bite.
A plain-English question — the same one the job ad describes — is translated to SQL by the agentic backend.
Curated cards run server-side against MotherDuck when eligible. The workspace separately labels any local inspection path.
Runs against my_db.fisher_demo.clinical_trial_visits. No synthetic numbers.
Each figure carries a falsifier — recomputed from the result set, not a stored number, so it can't quietly drift.
It's the role getting done: curated questions run live server-side against the warehouse; local inspection is labelled inside the workspace.
Open the live workspace →Provenance. Offline path: SYNTHESISED labelled 2,000-row clinical trial monitoring dataset (ITS-169, seed 20260613). Schema mirrors my_db.fisher_demo.clinical_trial_visits. ICH E6 (R2) GCP framework; MHRA inspection readiness context. NOT real trial data. Live path (dormant): server-side MotherDuck query against fisher_demo.clinical_trial_visits.
It's Sorted — I took ENSEK's job ads and didn't write a report on what AI could do. I built it. Get the rest sorted →
I'm trained on this proof and the real ENSEK: the Ignition meter-to-cash platform (seven modules), the move under Centrica in 2024, 7M+ energy accounts migrated for suppliers like British Gas and Utility Warehouse, and the Ofgem framing. Ask me how the Data Analyst function changes shape, or which open roles map to which Ignition module.