Own an AI Research Team

What Patrick personally does

Patrick:

• reviews the proposed workflow and whether an owned setup is appropriate;
• identifies the required inputs, outputs, decision points, and escalation points;
• agrees the pilot's acceptance criteria with you;
• configures and tests the environment;
• documents access, operation, review, and handover steps;
• runs the included workflow pilot;
• provides one correction round against the agreed criteria; and
• trains you or a nominated team member to operate the setup.

Patrick is accountable for the setup and services stated in the agreement. After handover, your nominated operator is responsible for running the environment and your lab or research group remains responsible for how its outputs are used.

One workflow pilot is included

The setup fee includes one pilot for one agreed workflow.

Before configuration begins, the scope records:

• the permitted input format and example inputs;
• the required output;
• who reviews or approves the output;
• the acceptance criteria;
• known exclusions and escalation conditions; and
• any privacy or infrastructure constraints.

Patrick runs the pilot and provides one correction round where the output does not meet the agreed criteria. A new workflow, materially different input, changed acceptance test, integration, or additional correction round is quoted separately.

Who owns what

You own and control the server account, your data, your research materials, and your lab or research group's decisions. You receive the project-specific configuration, workflow documentation, and other handover materials identified in the agreement.

Patrick retains ownership of pre-existing methods, templates, and general-purpose components unless the agreement transfers specific rights. Third-party software and model services remain subject to their own licences and terms.

The scope of work should identify any component that cannot be transferred, modified, or operated without an external service before you approve the setup.

Data flow is agreed before setup

A dedicated server can reduce mixing between projects, but "isolated" does not by itself establish that data never leaves the server. Data flow depends on the configuration, integrations, logging, backups, and model providers selected.

Before the pilot, Patrick and the client document:

1. what data the workflow receives;
2. where that data is stored and processed;
3. which people and services can access it;
4. whether any external provider receives prompts, files, metadata, or outputs;
5. the provider terms that apply;
6. retention, backup, and deletion requirements; and
7. what the workflow must do when restricted data or an unsupported request appears.

Patrick does not use client data to train his own models or reuse one client's confidential material for another client. If an acceptable data path cannot be configured, the workflow is redesigned or excluded.

Where an owned setup fits

Suitable

• A recurring literature-screening or evidence-extraction workflow with human review
• Repeated preparation of structured summaries from approved sources
• Routine transformation, validation, or reporting of research files
• Drafting or document workflows with a named human approver
• A lab process whose inputs, outputs, checks, and escalation points can be defined

Not suitable as a staff replacement

• Work where continuing human judgement is the main deliverable
• Participant care, clinical decisions, supervision, assessment, or disciplinary decisions
• Autonomous submission, publication, or spending; or bypassing institutional approval where it is required
• A workflow whose data use is not covered by consent, ethics, policy, or lawful authority
• A process with no accountable person available to review exceptions and outputs

The environment can support professors, students, RAs, research managers, and technical staff. It does not replace their academic, managerial, or institutional responsibility.

Pricing and recurring costs

Start at lab level with one scoped deliverable. Expand to additional workflows, labs, or a School & Department Plan only when useful.

Need one arrangement across several labs? See the School & Department Plan for deans and department heads.

Prices are stated in USD; HKD figures are approximate planning anchors.

The setup fee includes one workflow with agreed inputs, acceptance criteria, and one correction round. It does not include additional workflows, extensive custom integration, continuing operation by Patrick, or unlimited revisions.

Server and model charges continue for as long as you keep those services active. Actual cost depends on the selected provider, storage, usage, security requirements, backups, and model choice.

There is no automatic retainer. If you want Patrick to maintain or operate the environment after handover, the support scope, response terms, and fee are agreed separately.

Human responsibility remains in place

The setup can assist with preparation, checking, and repeatable production work. It does not take responsibility for research design, scientific interpretation, ethics, authorship, supervision, or institutional decisions.

Your lab approves the workflow, operators, and final outputs; your institution provides any policy, data-use, ethics, security, procurement, or deployment approval required in your setting. Patrick is responsible for the setup, pilot, documentation, and support expressly included in the agreement. Third-party services remain governed by their own availability, security, retention, and licensing terms.

Discuss a lab setup

Send the recurring workflow you want to support, who will operate it, the type of data involved, and any known infrastructure or policy requirements. Patrick will review whether an owned setup, a managed engagement, or a smaller trial analytics deliverable is the appropriate route before proposing scope.

Request a lab fit review