Services — Britt Biocomputing

Services

How I Work

Every engagement follows the same lifecycle — the depth scales with your risk.

CoU → Risk → Eval Design → Acceptance Criteria → HITL → Monitoring → Change Control

Assess

Find out where you stand

Before a regulator or auditor does.

$150

The Principle 8 Gap Screen

20 minutes. No pitch. No pressure.

In a single call, I'll tell you whether your current human oversight strategy would hold up under a Principle 8 audit. You'll leave with clarity on whether this is the right path — and if it is, what to do first.

Sprint

2–4 weeks

The "Human-on-the-Loop" Governance Sprint

You may not be ready to validate the model yet, but you need to validate the interaction. This sprint covers Stages 1–3 of the framework: defining the context of use, establishing risk-tiered requirements, and designing human oversight that actually works — not just a checkbox.

What you get

Context of Use statement — who uses this, for what decision, with how much autonomy

Human-AI interaction rubric — review thresholds, confidence routing, engagement requirements

Failure mode taxonomy — what happens when the model hallucinates, echoes, or agrees when it shouldn't

Best for: Teams building agentic workflows who need to prove they're in control before scaling. Product teams at software companies shipping AI features into regulated environments.

Inspection

The "Red Team" Inspection

A structured adversarial assessment of your human oversight controls. I simulate hallucinations, sycophantic outputs, and edge-case failures — then score exactly where the human-on-the-loop broke down.

What you get

Vulnerability scorecard — where oversight failed during simulated failures

Failure-specific recommendations — mapped to your context of use

Inspection-readiness gap analysis

Best for: Regulatory affairs teams preparing submissions. Quality/CSV/CSA teams doing inspection readiness. Clinical ops and CMC teams wanting a pre-inspection gap scan. IRBs or research compliance teams auditing AI workflows.

Build

Turn your AI from "we use it" into "we can defend it."

Full Build

6–10 weeks · Full validation for one context of use

Agentic Evidence Generation

The full build across Stages 4–5: evaluation that matches the risk, and an evidence pack that answers the questions regulators will actually ask. The goal is to move from "software validation" to evidence integrity — if the decision isn't traceable, it isn't evidence.

What you get

Decision lineage map — full traceability from model output through human review to final decision

Model credibility plan — the new "validation plan," built around your specific failure taxonomy and pre-registered acceptance criteria

Audit-ready evidence pack — context of use, risk assessment, test results, monitoring plan, and change control records in a single consolidated package

Best for: Clinical ops, safety, and regulatory teams who need to defend the output to an auditor. Software companies whose customers are in regulated industries. Academic medical centers and research institutes deploying AI for clinical decision support.

Training

5 virtual sessions (90 min each) + async templates

The GxP Bridge to AI Quality

For teams that already know GxP but need to translate that expertise to AI systems. This isn't "AI 101" — it's a structured bridge from traditional CSV/CSA thinking to the fit-for-purpose framework, covering how the principles your team already understands apply to models that learn, drift, and sometimes confidently hallucinate.

Best for: QA/CSV leads and data integrity officers who need to own AI quality but didn't come from a machine learning background.

Sustain

Validation isn't a one-time event

Things drift.

Retainer

Ongoing

Lifecycle Drift Assurance

Stage 6 of the framework — because the model was validated against a world that no longer exists the moment inputs change. New document types, updated SOPs, vendor switches, prompt engineering tweaks — any of these can silently degrade performance.

What you get

Quarterly drift monitoring report — documented proof that you're watching

Change control impact assessments — when someone prompt-engineers a new version, you have a documented evaluation of what changed and why it's still fit for purpose

Revalidation trigger log — pre-defined thresholds so you know exactly when "monitoring" becomes "revalidate"

Reviewer drift checks — periodic blind QA on human decisions against a gold standard

Best for: Organizations with LLMs in production who need their Principle 9 regulatory insurance policy.

Embedded & Fractional

For teams that need ongoing AI quality leadership

Not just a project.

Some organizations don't need a sprint or an evidence pack. They need someone embedded in the work — sitting in the architecture reviews, shaping the oversight design as features ship, and making sure the validation story holds together over time.

I work as a fractional AI quality and validation lead through Go Fractional, embedded part-time with your team on a retained basis.

This looks like

Joining product and engineering standups to flag validation and oversight gaps in real time

Owning the fit-for-purpose validation roadmap as your AI capabilities evolve

Serving as the internal voice on AI quality for regulatory, quality, and product teams

Building your team's capacity so this expertise lives inside your organization, not just with me

Ideal for grant-funded initiatives requiring specialized AI governance without the overhead of a full-time hire

Best for: Software companies building AI products for regulated industries. Life sciences organizations scaling from one AI use case to many. Teams that need a Head of AI Quality before they're ready to hire one full-time.

Not sure where to start?

I'll walk through your current setup, tell you where you are on the lifecycle, and recommend the smallest next step that creates the most clarity. No pitch. Just a map.

Book a 20-minute fit check →