AI integration basics

DreamApply provides access to AI-powered features through AI Highlighter. AI Highlighter integrates DreamApply with third-party Large Language Models (LLMs) and allows Administrators to perform admission tasks with the help of AI.

AI Highlighter can assist Administrators with the following tasks:

Data summarization: Generate concise summaries of application data
Document parsing and data extraction: Identify and extract key information from application data and uploaded documents
Classification and labeling: Analyze application data to automatically categorize and label applications based on specific criteria
Validation and consistency checking: Detect discrepancies or inconsistencies in applications
Text translation: Translate application materials into the institution’s primary language
AI-driven image verification: Automatically identify and analyze uploaded images, such as passport photos

Treat AI Highlighter as a helper, not a decision-maker. Administrators themselves remain responsible for interpreting AI responses and making final decisions. Always trace the AI reasoning trail that AI Highlighter displays for each AI output and verify it against the original documents. For details, see View AI responses.

LLM-based data processing

DreamApply does not develop or host its own AI models. Instead, it integrates with third-party LLMs to process application data.

At present, DreamApply relies on the following LLMs:

Category	Specification	Details
Models used	Google Gemini 2.5 Pro Google Gemini 2.5 Flash	Gemini models
Locations	Storage location: `europe-north1`, Finland Processing location: `europe-west4`, Netherlands	Deployments and locations
Data retention for API	No Google will not store any Customer Data as part of AI inference	Vertex AI and zero data retention
Trains on user data via API	No Google will not use Customer Data to train or fine-tune any AI/ML models without Customer’s prior permission or instruction	Service Specific Terms, Section 17. Training Restriction Sharing our data privacy commitments for the AI era

Prompt design and control

To process application data using AI, Administrators can configure prompts in DreamApply.

A prompt is an instruction in natural language that tells what the AI should do, what information to focus on and how to output the result. For example, a prompt may instruct the AI to summarize application data and identify the applicant’s strengths and weaknesses or to extract specific information from a transcript.

To guarantee that the output is accurate and relevant, Administrators can fine-tune the AI behavior using the following prompt settings in DreamApply:

AI model: Choose the AI model for data processing
Context materials: Add source documents and information to set the context for the admission task. AI Highlighter allows using the following context materials:
- Information submitted through the application form (including system section such as Scores)
- Documents uploaded through the application form and tasks (such as passport photos, transcripts and other)
Output structure: Define how results should be formatted or structured. AI Highlighter supports the following output modes:
- Unstructured, free-form text summaries: Can be helpful for generating high-level overviews and qualitative insights
- Data fields such as one-word labels, lists, dates, numbers and scores: Can be helpful for quantitative insights used for data filtering, ranking and analytics
Thinking budget: Control the AI internal reasoning process. The budget is defined in tokens, which are small pieces of information that the AI uses for processing and reasoning before generating a response. Increasing the budget provides longer and more detailed answers, but also uses more processing resources.
Temperature: Control the creativity or predictability of AI responses. Lower values make answers more precise and consistent and can be useful for data extraction or labeling tasks. Higher values allow more flexibility and are useful for summarization.

For the best output quality, follow the AI usage guidelines.

System prompt

In addition to Administrator-defined prompts, DreamApply uses a system prompt. The system prompt is a set of built-in instructions that define the overall AI Highlighter behavior and style. It applies to every AI request across the DreamApply system.

The system prompt provides guidance for the LLM to follow the internal DreamApply policies and rules. Administrators cannot edit the prompt directly through the DreamApply portal. However, they can contact the DreamApply team to request adjustments or customization of the AI behavior.

AI data flow

The following parties are involved in AI processing and handling application data:

DreamApply system: Collects and manages application data and shares the selected materials with third-party LLMs for AI processing.
Third-party LLM: Receives selected materials, processes them and returns outputs. The LLM is granted access only to the specific application data and documents selected in the prompt settings. The LLM cannot access any other materials or data within your account.

DreamApply never shares application data with third-party LLMs unless you explicitly run a prompt. DreamApply performs no background processing or automated data sharing.

When a prompt is run, AI Highlighter performs the following actions:

An Administrator selects one or more applications and runs a prompt to analyze their data. Alternatively, DreamApply triggers a prompt using rules.
DreamApply prepares the selected application materials and sends them to the chosen LLM over a secure communication channel.
The LLM processes the data based on the prompt instructions and generates the requested outputs.
The LLM returns the generated outputs to DreamApply. DreamApply then displays results on the AI highlights tab in the corresponding application card. Administrators can use this output for further analysis, data filtering and reporting.

Integration disclaimer

DreamApply provides AI functionalities through Google Vertex AI API. Use of AI functionalities is not mandatory, and institutions may use DreamApply without this integration.

The AI functionalities (for example, extracting data points, categorizing documents, summarizing applicant information, flagging incomplete fields and so on) are not developed or operated by DreamApply. By using these functionalities, you acknowledge and agree that:

DreamApply acts solely as an intermediary in providing access to the Google Gemini tools for application processing.
Google is responsible for the processing, security, and quality of the AI functionality itself, in accordance with its specific terms of service and data handling practices.

For details, see DreamApply’s AI Highlighter Special Terms.