Can users influence which sources DeepResearch uses or provide it with specific starting points for research?

Users cannot directly control or modify the core sources DeepResearch uses for generating results, as the system relies on predefined, curated datasets and algorithms to ensure reliability. However, developers and technical users can guide the research process by providing specific starting points or constraints. This is achieved through input parameters, APIs, or integrations that allow users to narrow the scope of queries or prioritize certain types of data.

For example, a user might submit a list of URLs, domain names, or document IDs alongside their query to focus the research on those sources. If DeepResearch offers an API, developers could structure requests to include source_filters or priority_domains parameters, instructing the system to weigh specific websites or databases more heavily. Similarly, if the tool supports custom document ingestion, users might upload proprietary datasets or research papers to be indexed and included in the analysis. These methods don’t alter the underlying source list but create a contextual layer that influences how the system retrieves and ranks information.

Technical implementation details matter here. For instance, a user-provided URL list might be processed through a pre-filtering step to exclude irrelevant domains or validate accessibility. If the system supports plugins or extensions, developers could build modules that inject custom data pipelines (e.g., scraping a private GitHub repo or querying a internal company database) before passing results to DeepResearch. However, there are limitations: rate limits, authentication requirements for private sources, or compatibility with the system’s data formats. These constraints ensure stability but require developers to structure inputs carefully. While users can’t edit the core source repository, they can steer the research effectively by combining targeted inputs with the system’s existing capabilities.