Income flow interpretation

A filing-derived financial read with key points and selected income-flow views.

Category: Financial
Forms: 10-K

Use cases#

Surface concise interpretation text alongside the underlying flow context.
Present filing-derived financial traits as structured key points.

Response schema#

Fields 28

Field path	Type	Description
`overall`# required	string	One-paragraph consolidated financial read (~15-25 words, plain text).
`key_points`# required	array items: string	Company-level financial-trait tags (~3-6 words each) readable from the income-flow figures, e.g. 'expanding operating margin'.
`income_flow`# required	object\| null	null when the input's `facts` carries no card-eligible key.
`income_flow.flow_summary`# required	string	One paragraph (~15-30 words) on the flow, grounded in the input's sankey summary (earned / spent / bottom).
`income_flow.items`# required	array items: object	At most one card per `fact_key`. Uniqueness is enforced after the loop, not here — JSON Schema cannot state it over an object array.
`income_flow.items[].fact_key`# required	string	Verbatim key into the input's `facts` object (e.g. 'cost:1', 'bottom_line').
`income_flow.items[].name`# required	string	English display name, copied verbatim from the fact's `label`.
`income_flow.items[].role`# required	string	This fact's character in the flow: 'revenue' (a top-line inflow), 'expense' (a cost/deduction), 'profit' (a positive bottom-line result), 'loss' (a negative bottom-line result), 'offset' (an elimination/contra item).
`income_flow.items[].key_points`# required	array items: string	Key Points
`income_flow.items[].core_summary`# required	string	One sentence (~15-25 words) on this fact's role, grounded in the fact's value / prior / yoy fields. Narrate year-over-year only from the input's own prior or yoy — never from a figure computed elsewhere.
`views`# required	array\| null items: object	null when no card-eligible view exists — including an input with no `views` key at all. Never an empty array.
`views[].id`# required	string	Copied verbatim from the matching input `views[].id`.
`views[].kind`# required	string	Copied verbatim from the matching input `views[].kind`.
`views[].label`# required	string	Copied verbatim from the matching input `views[].label`.
`views[].view_summary`# required	string	One paragraph (~15-30 words) on this view's breakdown, grounded in the view's sankey summary.
`views[].members`# required	array items: object	One card per entry in the matching input view's `breakdown.members`.
`views[].members[].member`# required	string	Copied verbatim from the input breakdown member's `member` — the XBRL member QName (e.g. 'cmtl:CommercialSolutionsSegmentMember').
`views[].members[].name`# required	string	Copied verbatim from the input breakdown member's `label`. A display label, so translations carry it; `member` is what joins this card back to the input.
`views[].members[].key_points`# required	array items: string	Key Points
`views[].members[].core_summary`# required	string	One sentence (~15-25 words) grounded in the member's value / prior / yoy.
`views[].members[].changes`# required	object	What moved for one breakdown member, each narrated only from that member's own prior / yoy. null when the input carries no basis.
`views[].members[].changes.new`# required	string\| null	New business or product (~6-15 words). A member with no prior figure is the candidate for this.
`views[].members[].changes.expansion`# required	string\| null	Growth item (~6-15 words).
`views[].members[].changes.risk`# required	string\| null	Deterioration item (~6-15 words).
`views[].members[].event`# required	object\| null	What happened to this member during the period — a launch, acquisition, plant, contract, regulation or reorganisation the narrative input states and names this member in. The only field this topic adds on top of the numeric card. `key_points`, `core_summary` and `changes` stay what they were — read from the member's own value / prior / yoy — so a consumer that already reads them keeps working. null unless the narrative input carries such a statement: standing facts (what the business consists of, how it sells, what the market is like) are not events, and neither is a figure the numeric fields already print.
`views[].members[].event.said`# required	string	One sentence (~15-25 words) restating what the quoted source text says.
`views[].members[].event.source_block`# required	string	Verbatim id of the narrative block this rests on (e.g. 'B12'), copied from the narrative input.
`views[].members[].event.source_quote`# required	array items: string	Verbatim fragments of that block, each a contiguous run of its text. A list rather than one sentence because the filing prints announcements as lists ('Second Quarter 2025:' / 'iPhone 16e'), and joining them into a single sentence is a rewrite, not a quote. Never merge, reorder or paraphrase them.

Example dataset#

An illustrative example for income_flow_interpretation. Use it to see the shape of the data before running a live request.

Top-level fields in the example dataset
Field	Summary
overall	Revenue grew while operating income remained positive for the reported period.
key_points	2 items
income_flow	null
views	null

Playground#

Response schema#

Fields 28

Field path	Type	Description
`overall`# required	string	One-paragraph consolidated financial read (~15-25 words, plain text).
`key_points`# required	array items: string	Company-level financial-trait tags (~3-6 words each) readable from the income-flow figures, e.g. 'expanding operating margin'.
`income_flow`# required	object\| null	null when the input's `facts` carries no card-eligible key.
`income_flow.flow_summary`# required	string	One paragraph (~15-30 words) on the flow, grounded in the input's sankey summary (earned / spent / bottom).
`income_flow.items`# required	array items: object	At most one card per `fact_key`. Uniqueness is enforced after the loop, not here — JSON Schema cannot state it over an object array.
`income_flow.items[].fact_key`# required	string	Verbatim key into the input's `facts` object (e.g. 'cost:1', 'bottom_line').
`income_flow.items[].name`# required	string	English display name, copied verbatim from the fact's `label`.
`income_flow.items[].role`# required	string	This fact's character in the flow: 'revenue' (a top-line inflow), 'expense' (a cost/deduction), 'profit' (a positive bottom-line result), 'loss' (a negative bottom-line result), 'offset' (an elimination/contra item).
`income_flow.items[].key_points`# required	array items: string	Key Points
`income_flow.items[].core_summary`# required	string	One sentence (~15-25 words) on this fact's role, grounded in the fact's value / prior / yoy fields. Narrate year-over-year only from the input's own prior or yoy — never from a figure computed elsewhere.
`views`# required	array\| null items: object	null when no card-eligible view exists — including an input with no `views` key at all. Never an empty array.
`views[].id`# required	string	Copied verbatim from the matching input `views[].id`.
`views[].kind`# required	string	Copied verbatim from the matching input `views[].kind`.
`views[].label`# required	string	Copied verbatim from the matching input `views[].label`.
`views[].view_summary`# required	string	One paragraph (~15-30 words) on this view's breakdown, grounded in the view's sankey summary.
`views[].members`# required	array items: object	One card per entry in the matching input view's `breakdown.members`.
`views[].members[].member`# required	string	Copied verbatim from the input breakdown member's `member` — the XBRL member QName (e.g. 'cmtl:CommercialSolutionsSegmentMember').
`views[].members[].name`# required	string	Copied verbatim from the input breakdown member's `label`. A display label, so translations carry it; `member` is what joins this card back to the input.
`views[].members[].key_points`# required	array items: string	Key Points
`views[].members[].core_summary`# required	string	One sentence (~15-25 words) grounded in the member's value / prior / yoy.
`views[].members[].changes`# required	object	What moved for one breakdown member, each narrated only from that member's own prior / yoy. null when the input carries no basis.
`views[].members[].changes.new`# required	string\| null	New business or product (~6-15 words). A member with no prior figure is the candidate for this.
`views[].members[].changes.expansion`# required	string\| null	Growth item (~6-15 words).
`views[].members[].changes.risk`# required	string\| null	Deterioration item (~6-15 words).
`views[].members[].event`# required	object\| null	What happened to this member during the period — a launch, acquisition, plant, contract, regulation or reorganisation the narrative input states and names this member in. The only field this topic adds on top of the numeric card. `key_points`, `core_summary` and `changes` stay what they were — read from the member's own value / prior / yoy — so a consumer that already reads them keeps working. null unless the narrative input carries such a statement: standing facts (what the business consists of, how it sells, what the market is like) are not events, and neither is a figure the numeric fields already print.
`views[].members[].event.said`# required	string	One sentence (~15-25 words) restating what the quoted source text says.
`views[].members[].event.source_block`# required	string	Verbatim id of the narrative block this rests on (e.g. 'B12'), copied from the narrative input.
`views[].members[].event.source_quote`# required	array items: string	Verbatim fragments of that block, each a contiguous run of its text. A list rather than one sentence because the filing prints announcements as lists ('Second Quarter 2025:' / 'iPhone 16e'), and joining them into a single sentence is a rewrite, not a quote. Never merge, reorder or paraphrase them.

Example dataset#

An illustrative example for income_flow_interpretation. Use it to see the shape of the data before running a live request.

Top-level fields in the example dataset
Field	Summary
overall	Revenue grew while operating income remained positive for the reported period.
key_points	2 items
income_flow	null
views	null