Knowledge Checker: fact check as you write

Knowledge Checker instantly scans your connected sources to strengthen claims, surface evidence, and catch contradictions before you share.

Hero image _ Knowledge Checker Gmail email mockup _ Address incorrect data point

Checks claims before you ask

Reads the claim behind your words, so support arrives before you go looking.

Backed by your sources

Pulls from apps, docs, and threads your team works in, so you get suggestions you can trust.

Keeps your information private

Works inside your existing permissions, so it only references information you have access to.

AI that gets your facts straight

Knowledge Checker is an agent that runs on Superhuman Go. It detects facts that need verifying, checks them against the sources your team already trusts, and provides the specific figure, quote, or document—inline, as you type.

Every claim checked, without you checking

As you write, a subtle underline marks any claim with evidence behind it. One click shows the source, another puts it in your draft.

Knowledge Checker inline rewrite suggestion

Look it up without looking away

Stats, timelines, and policy references get checked in the background as you write, so the manual fact-check and the tab-switching that came with it both disappear.

Knowledge Checker background source check card

Get the facts from your apps

Evidence comes from the docs, threads, and files your team works in, no matter how scattered, so your numbers match the source today.

Knowledge Checker connected apps

Why use Knowledge Checker

Proactive, not prompted

It checks your facts and strengthens your work without breaking your flow.

Meaning over keywords

It understands the claim you're making, not just the words you typed.

Names the source

Every card names its origin and why it matters, so you can judge the evidence before you use it.

Answers your follow-ups

Carry any detected claim into the Go side panel, so you can question it further or trace the source.

Broadens your search

It looks past your usual sources to find evidence you might've missed.

You control the sources

Add or remove connected knowledge sources any time from the agent settings page.

How to use Knowledge Checker

Install Knowledge Checker Agent

Step 1: Install the agent

Open the agent store in Superhuman Go and install Knowledge Checker.

Connect sources to Knowledge Checker

Step 2: Connect your sources

Choose your sources in Agent settings. Go connections carry over.

Knowledge Checker product example

Step 3: Start writing

Knowledge Checker monitors your draft, underlining claims with evidence.

Knowledge Checker addressing an incorrect data point

Step 4: Review and accept

Hover over the underline to see the source and why it matters, then insert — or open the Go side panel to go deeper.

Frequently asked questions

Knowledge Checker is an AI agent inside Superhuman Go that monitors your writing as it happens, detects factual claims, and surfaces supporting evidence, enrichment, or contradiction signals from your connected knowledge sources — without you leaving the page.

Search and chat wait for you to ask. Knowledge Checker doesn’t. It works continuously in the background and surfaces what belongs in your writing at the moment you write it — including evidence you didn’t know existed and wouldn’t have thought to look for.

Knowledge Checker currently connects to Superhuman Docs, the Microsoft stack, Google Workspace, Slack, and Atlassian. Any connectors you’ve already set up in Go are used automatically, and you manage them in your Account Settings.

Install Knowledge Checker from the agent store in Superhuman Go, then configure your connected data sources in Agent settings.

Knowledge Checker is available in beta on Pro, Business, and Enterprise plans.

It reads the assertion behind your words rather than matching keywords. When a statement makes a factual claim — a metric, a date, a customer outcome, a policy reference — it looks for the source in your connected knowledge that speaks directly to that claim.

No. Knowledge Checker surfaces evidence and flags conflicts. Nothing enters your document until you accept it.

Yes. Every inline card names the document, thread, or record behind it and explains why it’s relevant, so you can judge the source before you use it.

Only the sources you’ve connected to Superhuman Go — and within those, only content you already have access to. It can’t surface a document, thread, or record you couldn’t open yourself. Access is read-only, and every piece of evidence names the source it came from so you always know what you’re looking at. Nothing enters your document until you accept it.

Knowledge Checker follows Superhuman Go’s data practices. We don’t train third-party models on your data.

For Superhuman’s own models, Go respects your Customer Data Training Control settings. Access to your connected sources is read-only, and Knowledge Checker makes no new use of your connectors beyond the terms you agreed to when you first authorized them.

Learn more about Superhuman Go’s data and privacy policies.

Your reputation, backed by your trusted sources.