Understanding AI in DocsCheck
⚠ Not legal advice
No output from DocsCheck — including AI document scores, refusal risk indicators, questionnaire analysis, or AI-generated representation letters — constitutes immigration advice, legal advice, or a guarantee of any outcome. All professional decisions remain the sole responsibility of the instructed immigration professional.
What AI does in DocsCheck
DocsCheck uses OpenAI GPT-4o to power several platform features. Here is a precise description of what each AI feature does — and what it does not do.
Document analysis & scoring
AI analyses uploaded documents for quality (readability, resolution, completeness) and checks whether the document type matches what was expected for that visa route. Produces a quality score and flags issues.
Does not: assess legal sufficiency, confirm admissibility, or predict UKVI decisions.
Representation letter generation
AI drafts a representation letter using the client's questionnaire data, uploaded documents, and the visa route's typical evidence framework. Produces a draft for fee earner review.
Does not: provide legal advice, make legal arguments, or substitute for professional drafting. Every letter must be reviewed and approved by a qualified professional.
Message drafting
AI suggests draft responses to client messages based on the case details and a configurable knowledge base. Staff review and edit before sending.
Does not: provide legal guidance, advise clients on their immigration position, or send messages without staff approval.
Refusal risk indicators
Based on visa route requirements and the documents and questionnaire data present in the case, the platform flags gaps, inconsistencies, and missing evidence that are commonly cited in UKVI refusal notices.
Does not: predict whether an application will succeed or fail. UKVI decisions involve factors not visible to DocsCheck.
Parliament Bills analysis
AI analyses new UK Parliament Bills relating to immigration and produces plain-English summaries of potential impacts and suggested actions.
Does not: provide legal interpretation, advise on compliance, or substitute for legal counsel on legislative changes.
Client reminders
AI drafts personalised reminder messages to clients about outstanding documents or questionnaire completion, based on their case details.
Does not: send any message without staff configuration and approval of the reminder system.
How the AI is trained and what it knows
DocsCheck uses OpenAI's GPT-4o model via the OpenAI API. The model was trained on publicly available text data up to a knowledge cutoff date and is not specifically trained on DocsCheck data, your clients' data, or your firm's data.
Critically: the AI does not have access to current UKVI guidance, live Home Office policy updates, or real-time immigration law changes. AI-generated content should always be checked against current UKVI guidance, the Immigration Rules, and your professional knowledge of current Home Office practice.
We do not use your data to train AI models. Documents uploaded to DocsCheck, questionnaire responses, and client data are used only to provide the services described in these terms. They are never used to fine-tune, train, or improve AI models.
Your responsibilities when using AI features
As a regulated immigration professional using DocsCheck, you are responsible for:
- Reviewing all AI-generated outputs before relying on them for any professional purpose
- Verifying AI document scores against your own assessment — AI analysis may miss issues that a qualified professional would identify
- Editing and approving all AI-drafted letters before sending to clients or submitting to UKVI
- Checking all outputs against current UKVI guidance and the current Immigration Rules
- Applying your professional judgement to every case — DocsCheck output is a tool, not a decision
- Complying with OISC or SRA obligations — using DocsCheck does not modify or reduce your regulatory duties
Known limitations of AI in immigration contexts
- AI document analysis may not detect sophisticated document fraud
- AI-generated letters may contain inaccuracies or omissions that require correction by a qualified professional
- Risk indicators are based on structured rules, not on UKVI's actual decision-making processes, which are not publicly fully documented
- AI performance varies by document quality, language, and complexity of the client's circumstances
- The AI does not know about recent Home Office policy changes, country guidance updates, or tribunal decisions issued after its knowledge cutoff
Questions about AI in DocsCheck? Contact us at hello@docscheck.co.uk. We are happy to explain any aspect of how our AI features work before you sign up.