> ## Documentation Index > Fetch the complete documentation index at: https://docs.parley.so/llms.txt > Use this file to discover all available pages before exploring further. # Self-driving fields > What makes a field self-driving, how to design one, and what you get in return. A self-driving field is more than a place to store data. It's a structured signal that helps Parley understand what the record is, where it belongs, and what work may be relevant next. ## The four properties The field has a known type rather than an arbitrary string. See [Field types](/self-driving/field-types). The same field definition is used across records of the same object type. The field is attached to the correct Account, Contact, or Project. Agents and workflows can use the value to select context, identify missing information, prioritize work, or generate appropriate outputs. A Project field named `Case Status` is more useful than a paragraph saying "the case is currently pending." Because the value is structured, Parley recognizes the same concept across Projects and uses it consistently. Self-driving doesn't mean every field automatically performs an action. It means the field is designed to participate in the system — guiding organization, retrieval, reasoning, reporting, and future automation. ## Designing fields that self-drive well **A good field answers one stable question.** Prefer `Case Status`, `Target Date`, `Country of Citizenship`, `Current Immigration Status`, `Receipt Number`. Avoid fields that combine several questions or require prose parsing, like `Important Case Details` or `Status and Next Steps`. A field that needs to be read to be understood can't be filtered, compared, or acted on. **Use the narrowest useful type.** A date for a date. A number for a number. A boolean for a yes/no fact. A single-select for a controlled category. A collection for multiple labels. A note for narrative context. A file for source evidence. **Keep fields stable over time.** If a value changes frequently and the history matters, preserve the history in activity, notes, or source documents rather than overwriting the only record of what happened. ## What you get ### Better retrieval Typed fields help Parley distinguish relevant records from unrelated ones. A Project's status or target date becomes a reliable signal, without interpreting inconsistent prose. ### Better context selection When an agent works on a Project, the linked Accounts, Contacts, fields, notes, and files form a coherent context. The field tells the agent what kind of work it's handling; the links tell it who and what the work concerns. ### Better completeness checks A structured field makes missing information visible. A blank target date, a missing status, an absent country of citizenship — these become identifiable data gaps rather than something overlooked inside a paragraph. ### Better reporting and filtering Consistent types and controlled options make it possible to group and compare records. This is why `Filed` shouldn't sometimes be entered as `filed`, `Submitted`, or "sent to USCIS." If the concepts differ, use separate fields or options. If they're the same, use one controlled value. ### Better work products Agents use field values alongside files and notes when producing drafts, forms, letters, and packets. The field supplies structured facts, the files supply evidence, the notes supply operational interpretation. ## Capturing facts once By capturing each fact once, in the correct object and the correct type, Parley can organize context, identify gaps, support retrieval, guide workflows, and help agents produce better work with less manual reconstruction. See [Where information belongs](/self-driving/where-information-belongs) for how to decide, and [Case management](/self-driving/overview) for how Parley keeps these fields current on its own.