The tools

Specialized tools inside general-purpose AI chat.

Ledger is a general-purpose AI chat over your agency's own records. Inside it sit specialized tools — FOIA redaction, map-based intelligence, device and media search — that open when the work calls for them. Everything they do is cited, and everything is logged.

Summary table

Tool	Kind	What it does	Who uses it	What the person decides
Chat with your records	AI chatAsk, draft, summarize, compare	Ask questions of reports, carrier returns and case files. Draft, summarize, compare.	Detectives, records, command staff	Every claim cites a page or a row.
FOIA redaction	Specialized toolExemption-cited redactions	Proposed redactions on a records packet, each under the exemption it cites.	Records staff	Accept or reject each one, with a reason.
Map-based intelligence	Specialized toolRecords and phone data on one map	Report addresses and phone data on one map, filtered and asked in plain language.	Detectives, analysts	What a record can and cannot prove is said out loud.
Device and media	Specialized toolSearch inside the warrant's scope	A phone extraction searched by description, inside the warrant's scope.	Digital forensics	A person confirms every finding.
Everyday tasks	AI chatSame chat, same log	Elements of the crime, checklists, rewrites, a calculation over a spreadsheet.	Anyone with a login	Same log, same boundary.

Chat with your records

AI chat

Ask questions of reports, carrier returns and case files. Draft, summarize, compare.

Who uses it: Detectives, records, command staff
The decision: Every claim cites a page or a row.
What it will not do: It does not answer without showing the page or row behind the answer.

Ledger's chat tool answering a question across a fictional case's records, every claim carrying a numbered citation to the row or page it came from. — The chat tool answering a question across eleven reports. Each sentence carries the page it came from.

FOIA redaction

Specialized tool

Proposed redactions on a records packet, each under the exemption it cites.

Who uses it: Records staff
The decision: Accept or reject each one, with a reason.
What it will not do: It does not release anything on its own.

The redaction reviewer on page four of a fictional records packet: proposed boxes on the page, each listed on the right under the exemption it cites, awaiting a person's accept or reject. — A records packet with proposed redactions. Each one names the exemption it is claimed under.

Map-based intelligence

Specialized tool

Report addresses and phone data on one map, filtered and asked in plain language.

Who uses it: Detectives, analysts
The decision: What a record can and cannot prove is said out loud.
What it will not do: It does not place a person at a point from a cell record.

The map tool showing a fictional carrier return over a two-hour window: serving cell sites, handset records and one incident, with the answer written out beside it. — The map tool with a carrier return plotted and a filter applied. Every plotted point keeps its source row.

Device and media

Specialized tool

A phone extraction searched by description, inside the warrant's scope.

Who uses it: Digital forensics
The decision: A person confirms every finding.
What it will not do: It never searches outside the warrant scope.

The evidence grid: eight image results from a fictional phone extraction, each with a match score and a stated reason, waiting for an examiner to confirm. — A phone extraction searched by description. The warrant's scope is shown above every result.

Everyday tasks

AI chat

Elements of the crime, checklists, rewrites, a calculation over a spreadsheet.

Who uses it: Anyone with a login
The decision: Same log, same boundary.
What it will not do: It does not make a decision about any person.

The same general-purpose chat surface used for everyday work on a fictional case, with citations under each statement. — An everyday task: the elements of a charge, turned into a checklist a detective can work.