Portrait

User GuideScreening

Portrait Textual Reasoning

Search Over Company Documents, Portrait Primers, and Portrait Case Studies

A qualitative filter searches company documents — along with Portrait Primers and Case Studies — and surfaces the data points or situations that you care about most. Remember to use full sentences and provide tangible details (similar to how you would give instructions to a new member of your team).

Based on what you write in the query box, our models will expand your prompt into a series of keyword-like searches to find relevant paragraphs. If a paragraph appears to suggest that a Company matches your prompt, our system returns the result and explains its rationale and cites its source in the final output.

The textual reasoning filter with the plain-language query box and evidence window The textual reasoning filter with the plain-language query box and evidence window

Because Portrait Screens can run at intervals (weekly, monthly, quarterly), we recommend that users use "Relative" date ranges to ensure that incremental screening runs capture new information.

Background on Portrait Case Studies and Portrait Primers

Portrait has applied deep reasoning to generate primers and case studies across a wide range of companies. Every qualitative filter includes these alongside the underlying company documents, giving the system story-level context that a simple data point search cannot provide — powerful for identifying "stories" that fit investment frameworks, or event-driven scenarios like management change, M&A, or shareholder activism.

Screening Common Use Cases

Finding new ideas that fit your mental models or investment frameworks, or have interesting catalysts

Finding companies exposed to a key development or factor

Picking up important datapoints or signposts within an ecosystem

Screening Prompting Tips

  • Think of screening as giving directions to an eager Analyst that doesn't need to eat or sleep - the system responds best to a clear sentence or command.
  • Write the prompt such that an intern just reading a source document could ascertain relevance
  • Bad example: "Find me companies with red flags"
  • Good example: "Find me companies that have recently replaced their auditor, and the new auditor is not KPMG, E&Y, PwC, or Deloitte"
  • The screener will lean towards errors of commission (i.e. showing you results that could be relevant) vs omission
  • For more in-depth screening on multiple quantitative + qualitative filters, email requests@portraitresearch.com