Submission-ready pharmacometrics, on a fixed timeline.

No, and the difference is structural rather than cosmetic. A CRO sells time, so the incentive runs toward more hours. We sell a defined deliverable at a fixed price, so the incentive runs toward finishing it well and quickly.

The practical consequence is that the senior scientist who scopes your project is the one who does the modelling. There is no pyramid underneath, and nobody is learning pharmacometrics on your programme.

A fixed price agreed before any code is written, against a scope agreed in the same conversation. The kickoff price is the final invoice.

The number depends on the question. A focused non-compartmental analysis is a modest, well-bounded piece of work. A population model with covariate analysis and a submission-ready report is considerably more. Tell us the question and you get a number and a description of exactly what it buys.

That risk sits with us, not with you. It is the point of fixed pricing. If a model resists convergence, or the covariate structure is more tangled than the data first suggested, we absorb the extra work.

What we will not do is quietly narrow the question to fit the budget. If the honest answer changes, you hear it while there is still time to act on it.

Agents do the slow, mechanical parts: assembling and checking datasets, generating diagnostic plots, drafting boilerplate report sections, and cross-checking outputs against the source data. They run in a controlled environment, not a public AI service, and your data is never used to train anything.

Every scientific judgement is made by a pharmacometrician. Model structure, covariate inclusion, what the diagnostics actually mean, and what goes in front of a regulator are decided by a person who signs their name to it. The agents make the work faster; they do not make the decisions.

Yes. Every engagement hands over the analysis dataset, the code, the model files, and a report with its tables and figures. Everything is generated from code, so it can be re-run, audited and extended.

Where the choice is ours we favour open tools, precisely so you are not dependent on a licence, or on us, to reopen your own work in two years.

Usually not one that prevents the work, though it changes what can be concluded. Sparse sampling still supports a population model; that is part of why the method exists.

What matters far more is that dosing and sampling times are reliable. Where they are not, we will tell you what can and cannot be concluded rather than papering over it. On real programmes a large share of the timeline goes into assembling and checking data rather than modelling, and we say so up front so the schedule is honest rather than optimistic.

Yes. We prepare submission-ready population PK and exposure-response analyses documented to the standard FDA, EMA and PMDA reviewers expect, including the material an IND, NDA or MAA requires. We also help draft briefing documents and responses to questions that come back.

One honest caveat: acceptance depends on the whole dossier and the specific question, not on model quality alone. What we commit to is work that is documented, reproducible, and presented with its assumptions and limitations stated plainly.

Earlier than most teams do. The highest-value moment is usually before a study is locked, when the design can still be changed so it produces data a model can actually use. The second best is as soon as a readout lands.

You do not need a modelling strategy to get in touch, only the question. If modelling is the wrong instrument for it, we will say so on the call rather than after you have paid for it.

We sign NDAs as a matter of course, and will sign yours before reviewing any non-public material. Client results are never published without explicit written permission.

Analyses run in an isolated, access-controlled environment. Data is not sent to public AI services, is not retained beyond the engagement without agreement, and is not used to train models.