Skip to main content

Files

Every file a stage produces and every file you upload is held in the project's document store. Slate reads that store, and the store is also readable outside Slate. See Slate Vault for the store itself.

File types

Each file carries one type. The type drives which stage finder shows it and which folder it appears in.

TypeHolds
SpecificationA product specification
Market ReportMarket research and gap reports
VOC ReportPain point reports and review summaries
DatasetSpreadsheets and normalised report data
CopyListing copy versions
Marketing ImageGenerated product images
ManualUser manuals
Brand AssetBrand guideline documents and reference images
PositioningPositioning analyses
Profit ModelPrice point and profit reports
Factory BriefFindings written for the factory
Ops ReportReports produced by scheduled jobs
MiscAnything else

Uploading

Files are uploaded from the file rail of any stage page, from the project creation form, or by dropping them onto a conversation.

  1. Drag the files onto the upload area, or select it to choose files.
  2. Slate detects each file's type and shows it.
  3. Confirm or correct the type.
ConstraintDetail
Maximum file size25 MB
Readable formatsPDF, Excel, Word, CSV, Markdown, plain text and images
PDF readingThe first 50 pages
Spreadsheet readingThe first 200 rows and 40 columns of each sheet
Text readingThe first 200,000 characters

A file larger than the cap, or in a format Slate cannot read, is still stored. Only its contents are unavailable to the model. Truncation is stated on the file.

Finding a file

From a stage

The file rail on every stage page carries a finder pre filtered to that stage's file types. Set TYPE to widen it, or search by name. Selecting a file attaches it as input to the stage.

From the sidebar

Expanding a project reveals Market Reports, Specifications and Assets. Each opens a folder view.

ControlEffect
Type filterRestricts the list to one file type
Generation filterRestricts the list to one generation
Search filesMatches the file name
List and GridSwitches the layout

Reading a file

Open a file to read its contents beside its details.

DetailMeaning
TypeThe file type
VersionThe version number of this file
StageThe stage that produced it
GenerationThe generation it belongs to
Git commitThe recorded change it arrived in
Created and UpdatedTimestamps
PathIts location in the document store
Document propertiesAny metadata fields written into the file

Markdown files render as formatted text. Spreadsheets and CSV files render as a table, with Raw and Table switching the view. Long tables show the first rows and say how many are shown. A file type Slate cannot render reads Preview not available for this file type.

Select Download to save the original file.

Comments

Comments hold review notes left by the model and replies left by your team.

  1. Open the file.
  2. Enter the comment.
  3. Select Post.

A comment may be anchored to a point in the document. Model authored notes are bylined with the model that wrote them. Members with the viewer project role can read comments and cannot post them.

Versions

Slate writes a new version rather than overwriting a file. Earlier versions remain readable. The version number appears against the file in every list.

Parsing an Amazon report

The report parser turns an Amazon Search Term or Business report into a structured dataset, with every row traceable back to its source row.

  1. Open Operations on the project.
  2. Select Choose report.
  3. Select Parse.
ConstraintDetail
Formats.xlsx and .csv
Maximum size25 MB
Row capReached files are marked Truncated
PermissionOperator project role or above

The result names the report type it recognised, the rows kept, the rows skipped and the dataset written. An unrecognised report reads Unrecognised report type. Select Open dataset to read the output.

Confidentiality

Uploaded review and returns exports carry customer data as supplied. Reports Slate produces from them are written from anonymised data: reviewer names and profile addresses are hashed at ingestion and profile images are discarded. Uploaded originals keep whatever they contained.