Pilot programme
One candidate. One question. One measurable decision.
A pilot is the smallest useful test of the loop. It is designed around one candidate, one resistance context and one decision the sponsor needs to make. What follows is the standard shape; the details are scoped with each sponsor.
Method
Write the decision down first.
Before any experiment is chosen, the pilot records the decision it should inform and the go/no-go criteria. That keeps the batch small and the result readable, including when the answer is “stop”.
Pilot shape
Eight deliverables.
What every pilot includes.
01
Decision brief
The question, the constraints, the budget and the decision the pilot should inform.
02
Ranked batch
A small set of combination experiments, each with a reason it was chosen.
03
Uncertainty map
Where the ranking is confident, where it is not, and what we would decline to rank.
04
Lab coordination
Assays run on authenticated models with a qualified laboratory partner.
05
QC & model update
Readouts checked for quality, negatives included, then used to update the next plan.
06
Independent confirmation
The most decision-relevant signals re-tested out of batch or with an orthogonal readout.
07
Executive package
What to test next, what reproduced, what to stop, in plain language.
08
Audit trail
A record of each choice, the data behind it and what changed it.
If it works
Four possible horizons.
An earned path, not a promise. Each step only makes sense if the one before it was worth it.
- H1
One decision
A single pilot wave on one candidate and one resistance question.
- H2
Repeated waves
Further waves on the same programme, each shaped by what the last one learned.
- H3
More resistance contexts
The same approach applied to other candidates or settings.
- H4
A learning layer
Lessons carried across a sponsor’s own programmes, within that sponsor’s governance terms.
Scope and cost are agreed per pilot. Laboratory work runs with qualified partners; Cytogent is not a competing wet lab.
