AI in Feith · Conversational

Chat with your records, and only your records.

Ask Me answers from the documents in your Feith instance, and only those. Unlike general AI tools that draw on the open web, every answer is grounded in records you hold and cited to the source, so it's traceable, not generated.

Request a walkthrough All six capabilities ›

0 Records copied out to answerYour model reads them inside your boundary

0 Answers from the open internetGrounded only in the records you hold

1 Permission model, inheritedIt sees exactly what the signed-in user sees

Every Answer cites its sourceClick a citation straight through to the record

The bigger picture

Your records already tell a story. Ask Me helps you hear it.

Across thousands of emails, memos, and case files runs a thread no one has time to follow: who decided what, when, and why. That story is already there, written into the record. Ask Me lets you ask for it in plain language and hear it back, every line traced to the document it came from.

The problem

A general chatbot doesn't know your records.

Teams already ask AI to summarize, find, and explain. But pointing a public tool at agency work goes wrong three ways at once, and none of them are defensible in a federal record.

Open-web answers aren't your records

General chatbots draw from the public internet. They can sound authoritative while citing nothing you own, or invent the answer outright.

Exports copy records out

Bolting AI onto an exported copy moves records into a second system, and out of the controls that protect them.

They ignore who may see what

A general tool has no notion of your permission model, so it can surface a record to someone who was never cleared to read it.

You need AI that reads only the records in front of it, sees only what the user is authorized to see, and cites every answer back to its source.

What it does

Answers grounded in records you can open.

Feith centralizes every record (emails, chat logs, documents, images, and audio) into one governed repository that's ready for AI. Ask Me reads that repository and nothing else: ask a question in plain language and it answers from the records you hold, with a citation on every claim.

Permission-aware retrieval: Ask Me runs with the signed-in user's permissions and sees only what they're allowed to see.
A citation on every answer: each claim links to its source record, and the document ID opens it straight in Feith.
Ask from anywhere: right-click Chat on a folder, a search result, or a saved Collection and start asking. No setup, no model tuning.
Bring your own model: run Ask Me on GPT, Claude, an on-prem open model, or whatever your agency has authorized.

The answer is grounded in records your team has verified, not in someone else's blog post. Every claim cites a record you can open.

Worked example

Question

“Summarize our telework policy changes since 2023.”

Ask Me permission-aware

1,204records in scope

Permission filter · runs with your permissions

7authorized records

Records outside the user's authorization never enter the context window.

Cited answer

A grounded summary: every claim traces to a record you can open in Feith.

HR-2023-114 HR-2024-061 MEMO-2025-08

How it works

It reads only your records, and only the ones you're cleared for.

Ask Me runs with the signed-in user's permissions, retrieves only the records that user is authorized to see, and reads them with a model you've approved running inside the same boundary. Nothing is exported; the answer comes back cited, and every question is written to the same audit trail your records already carry.

You ask in plain language

Ask Me

Runs with your permissions · retrieves only authorized records

Your model reads them, inside your boundary (bring your own)

Cited answer: every claim links to its record

Ask Me pipeline, top to bottom: you ask in plain language; inside Ask Me it runs with your permissions and retrieves only your authorized records, with permission and retrieval working together, and your model reads them inside your boundary; then a cited answer comes back, every claim linked to its record.

Where you ask it

Right-click Chat, wherever the records are.

Ask Me meets the records where they already live: no new screen to learn, no collection to build first.

From a folder

Right-click any folder and choose Chat to ask across everything inside it. No export, no setup.

From a search result

Run a search, then ask Ask Me about exactly the set it returned. Narrow the scope, then question it.

From a saved Collection

Save a working set as a Collection and chat with it, then run AI Redaction across the same collection.

See AI Redaction ›

In practice

What people ask it.

The same chat, pointed at a different set of records. A few of the questions teams put to it:

Explore policies and SOPs

“What changed in our records-retention schedule this year?”

Grounded answers, each cited to the source.

Prepare FOIA responses

“Show all records released for Project Orion in FY24.”

FOIA staff assemble release packages from cited, authorized records.

Contract research

“What's the renewal date and termination clause for Vendor X?”

Critical terms located, with the clause cited so you can verify it.

IT and security support

“Have we seen this error code before?”

Relevant past issues and fixes surface fast.

Explore the rest of AI in Feith

Five more capabilities, each its own deep-dive.

These are the other capabilities in the AI in Feith family — separate tools that each share the same permission model, the same audit trail, and your choice of large language model.

Back to AI in Feith ›

Get started

Every authorized user, an instant expert in your records.

Bring a collection of records and a question your team asks every week. We'll answer it live, with citations you can click straight through to the source.