Toggy hides posts. It never deletes them. Anything Toggy hides stays in your community and can be restored with one click.

How Toggy moderates content

When a member publishes a post or reply, Bettermode sends it to Toggy. Toggy runs the content through a pipeline that completes in seconds:

1. Thread context. For replies, Toggy walks up the reply chain and reads the parent post. A blunt reply in a heated thread reads differently from the same words in isolation.

2. Author history. Toggy checks how this author's earlier content was decided.

3. Classification. Toggy scores the content against different categories and returns a confidence score for each one, with a written explanation.

4. Rules. Toggy compares each score against that category's threshold. Categories at or above their threshold match; categories below it do not.

5. Enforcement. Toggy combines what the matched rules ask for. If any matched rule hides, Toggy hides the post in your community. If any matched rule flags, the decision is recorded for review.

6. Evidence and audit. Toggy stores the scores, the rules it evaluated, its reasoning, and what it did.

The Agent page shows each step as it completes.

Decisions Toggy can make

Toggy produces one of three decisions. The decision is a description of what Toggy actually did, so it follows directly from the matched rules' Hide and Flag settings.

What Toggy checks for

Toggy scores content against twelve categories. Each one has a safety tier that governs how much your community profile can soften it, and a default action and threshold that come from your sensitivity setting.

The tier decides what your own settings can change:

• Soft categories are waivable. You can turn them off for the whole community or exempt individual members through the Whitelist.

• Severe categories are subject to the safety floor described below. Your community profile cannot talk Toggy out of them.

• Standard categories are neither waivable nor floored.

The safety floor

Whenever Toggy's own score for a severe category reaches 50%, it flags the post for review. This happens even when the score sits below the threshold you configured, and even when your community profile treats the behaviour as normal.

The floor only ever adds review. It never hides a post on its own. A severe category still has to cross its configured threshold before Toggy hides anything.

Setting up Toggy

1. Install the Toggy app from your community's App Store. You need an Admin account.

2. Open the Toggy dashboard and sign in with your Bettermode email. Toggy lists the communities where the app is installed. Select the one you want to moderate.

3. Go thought onboarding flow.

Bettermode connects Toggy to your community when the app is installed.

Tuning what Toggy enforces

Go to Settings → Rules. Sensitivity is the top-level control, and it decides where every rule's threshold and action come from.

Conservative, Moderate, and Aggressive are presets. Each one sets a threshold and an action for all twelve categories, and it overrides any per-rule edits you have saved. Conservative demands more confidence before Toggy acts. Aggressive demands less. Hate Speech, for example, is 85% on Conservative, 75% on Moderate, and 65% on Aggressive.

Custom uses your own per-rule settings. Change any rule's threshold, action, or on/off switch and your community moves to Custom automatically, so your edit takes effect.

Thresholds are clamped to a 50–95% range. A threshold below 50% would never fire, because 50% is also the score a category needs before Toggy considers it at all.

The rest of Settings covers the surrounding controls:

• Spaces — the spaces Toggy watches.

• Profile — your community's tone, norms, topics, and boundary examples.

• Learned Norms — what Toggy has worked out from your own moderation decisions. It refreshes nightly and on demand.

• Whitelist — per-member exemptions from soft categories.

• Blocked Users — members whose posts Toggy always hides.

• User Moderation — scanning member profiles for spam links.

• Agent — the run state and the circuit breaker.

Reviewing Toggy's decisions

Agent is the live feed. Cards appear as content arrives and animate through each pipeline step. The header shows four running totals: automation rate, posts processed, average processing time, and override rate. Click any card to see the reasoning, the per-category scores, the rules Toggy evaluated, the thread context, and the author.

Review Queue is the moderation inbox, with six tabs:

• All — every decision.

• Flagged — flagged items that still need a human decision.

• Skipped — items that were not analyzed but can be manually re-queued if needed

• Reported — everything Toggy flagged.

• Removed — everything Toggy hid.

• Appeals — decisions a member has disputed.

• Resolved — items a moderator has already settled.

An item that Toggy could not classify — because the model refused, or the community ran out of credits — is recorded as flagged and shown as Unmoderated. It waits in the queue for a human rather than being judged.

Overriding a decision

Open any decision and choose Approve or Reject. Toggy shows its own analysis of what it decided and where it went wrong, suggests reasons, and previews which thresholds your override will move. Confirm, and the decision updates everywhere at once. Approving un-hides the post. Rejecting hides it. The audit log records who acted and why.

You can do the same from the Review Queue, where you can restore a removed post or hide an approved one, with your reasoning attached.

Each override also tunes the rule behind it:

• You approved something Toggy hid, so the threshold rises and Toggy acts less often on similar content.

• You rejected something Toggy approved, so the threshold falls and Toggy catches similar content sooner.

• You resolved a flagged item, so the threshold moves by half as much.

The first adjustment moves a threshold by 5 points. Each later override moves it slightly less, down to a floor of 1 point, so no single correction swings a rule. Adjusted thresholds stay within the same 50–95% range.

Overrides are stored as per-rule settings, so they take effect once your community is on Custom sensitivity. On a preset, your corrections are recorded but the preset's thresholds continue to apply.

If you hide or un-hide a post directly in Bettermode, Toggy records that as a moderator override too.

Appeals

A member whose content Toggy restricted can open a Statement of Reasons: a page explaining what was decided and why, in plain language, with no internal rule ids or model details. From there they can appeal.

Open appeals land in the Appeals tab of the Review Queue with a deadline attached. The default window is 7 days.

Ruling on an appeal sets the decision's status to Restored if you accept it, or Appeal resolved if the original decision stands.

Controlling the agent

Running or Stopped. Use the control in the dashboard header, or Settings → Agent. Running means Toggy moderates new posts and comments as they arrive. Stopped means Toggy ignores everything incoming. Content published while Toggy is stopped is not moderated afterwards.

Credits

Toggy spends one credit per post or comment it moderates, and one per member profile it scans. Learning from your own decisions is free.

Starter includes 2,000 credits a month and Professional includes 10,000. You can buy more at any time from Billing.

When a community has no active subscription or runs out of credits, Toggy stops judging content. Incoming posts go to the Review Queue as Unmoderated and stay visible. Nothing is hidden on an empty balance, and you are not charged for an attempt that produced no decision.

Analytics and audit logs

Analytics has three tabs. Overview shows items processed, automation rate, average processing time, and violation rate. Accuracy shows accuracy, override rate, and feedback collected. ROI estimates the moderator time and cost Toggy has saved.

Audit Logs lists every action Toggy or a moderator took: each decision, each override, each appeal ruling, and each hide or un-hide mirrored back from Bettermode. Every entry records who acted, when, and why.

Ask ChatGPT or Claude which platform to use and you get a confident answer with a short list of vendors on it. What feeds that answer is mostly customer threads, reviews and public Q&A, written by people who don't work at the company being recommended.

That puts customer marketing and community teams in a strange spot. The advocacy work you already do, the references and quotes and customer stories, is the raw material for how a buyer finds you now. Most of it is sitting in a slide or a gated PDF where no assistant will ever read it.

We're doing a session on this on Thursday, September 24 at 12:00 pm ET. Ami Asadi, our Chief Product Officer, and Jeff Ernst, CEO and co-founder of SlapFive, work through how the two programs connect. RSVP below, and the recording goes out afterward if you can't make it live.

Agenda

Where AI assistants pull customer evidence from

The kinds of pages answer engines cite when they recommend a vendor, and a quick way to check whether any of yours are in that set. Most teams have never looked.

The advocacy to community loop

How references and customer stories become public, indexable conversations. The hard part isn't the publishing. It's doing it without burning out the same dozen advocates who say yes to everything.

A publishing rhythm for customer proof

What to publish where so search engines and answer engines both pick it up, and how advocacy and community end up sharing one calendar instead of running two.

We shipped a lot this summer, and the one I keep coming back to is this: your community is now something Claude and other AI assistants can read and act on through MCP.

Here's what's new, and why each piece matters.

AI tool connectivity through MCP

Your community now speaks MCP, the standard interface AI assistants use to read and act on external systems. Point Claude or another assistant at your community and it can pull what it needs, or take action, with no custom integration first. We'll demo live workflows on the call.

A first look at community agents

The Moderation Agent is live in beta today, working the queue with you. Two more are coming: a Community Builder Agent that sets up a community through conversation, and an Engagement Agent that surfaces the members worth re-engaging before they go quiet.

HubSpot 360, now in beta

Community engagement lands directly on your HubSpot contacts, companies, and deals, so member activity sits next to email opens and deal stages. The CSV export you've kept to answer “is the community doing anything” can go away.

Salesforce 360, now in beta

Posts, joins, replies, RSVPs and attendance flow into Salesforce as three native objects, and reps see community activity right on the contact record. Profile fields sync back the other way.

Analytics 2.0

Analytics rebuilt from the ground up for speed and accuracy across members, content, and engagement, and the foundation for tying engagement to pipeline. We'll show where that stands on the 27th.

What's next

This release is mostly connective tissue, getting community data to where decisions already get made. The agent layer is where we go next, and the foundation we laid this summer is what makes it possible.

You'll need admin access to Bettermode and a HubSpot user who can install apps (Super Admin, or the App Marketplace Access permission).

Part 1: Connect HubSpot

1. In your Bettermode Administration view, go to App store → Apps & Integrations.

2. Click HubSpot 360, then click Install. Contact the Bettermode team if you don't see the app.

3. In the app's settings, click Connect HubSpot. You'll be redirected to HubSpot.

4. Log in, pick the portal you want to connect if your user has more than one, review the requested permissions, and click Connect app.

5. You're redirected back to the HubSpot 360 settings page. That's the whole connection! The sync, ticket, field mapping, and Knowledge Base options are now unlocked.

💡 If you have a HubSpot test portal, connect that first and validate everything before connecting production. Disconnecting and reconnecting to a different portal takes a minute.

Part 2: Turn on sync

1. On the settings page, switch on Enable sync.

2. The first time you enable it, HubSpot 360 runs a one-time backfill of your existing members. You can watch the progress on the settings page; how long it takes depends on the size of your community.

3. From then on, member changes reach HubSpot within seconds, and a reconciliation pass runs every half hour to catch anything a missed event left behind.

What lands in HubSpot

Members sync as standard Contacts, matched by email and associated to existing Companies by email domain. Each synced contact carries six properties in a Bettermode property group: member ID, member name, joined at, last seen at, spaces, and badges.

To see them, open any contact and search the property list for "bettermode", or filter a contact view by Bettermode member ID is known to list every synced member.

Part 3: Map member fields to contact properties

A field mapping is a link between one member profile field in Bettermode and one contact property in HubSpot. Once two fields are linked, HubSpot 360 keeps them in sync automatically: when the value changes in the source system, the other side gets updated on the next sync.

1. On the settings page, find Field mappings and click Add mapping.

2. Pick a Bettermode member field on one side and a HubSpot contact property on the other.

3. Choose the direction. Each mapping is one-directional: Bettermode → HubSpot or HubSpot → Bettermode, decided per field.

4. Click Save. New mappings backfill existing members on the next pass, so you don't need to re-run anything.

Part 4: Create HubSpot tickets from posts

Some community questions need to become support tickets. This puts a Create a HubSpot ticket option in the (…) menu of every post (visible to staff only).

1. On the settings page, switch on Enable ticket creation.

2. Navigate to any post, click the (…) menu at the top right, and click Create a HubSpot ticket.

3. Fill in the ticket details. You can search your HubSpot contacts to attach the right one.

4. Click Create ticket. The ticket is created in HubSpot with the post's context attached.

Part 5: Search your Knowledge Base from the community

If your portal has a HubSpot Knowledge Base (Service Hub Professional or Enterprise), federated search can surface its articles inside community search.

1. On the settings page, switch on Enable Knowledge Base search.

2. Search the community for a phrase you know appears in an article. Knowledge Base results appear alongside community results.