intelligencesupport

Agent catalog

The catalog lists the thirteen standard agents, open to everyone to view and add to their agents, and the tailor-made agents BearingBridge builds on request.

For
Everyone
In the app
/agents/catalog
Updated

Watch the video

Put an agent to work from the catalog

The agent catalog lists every agent you can use. Standard agents are open to everyone: you read one in full and add it to your own agents, without asking anybody first. Tailor-made agents are built by BearingBridge on request. To open it, click Agents in the header, then the Catalog tab.

Standard agents

Standard agents come with the app. Each reads a fixed set of data on a schedule, and none of them runs until it is added to someone’s agents.

Each card shows the agent’s catalog code, its name, what it does, and the data it reads, on a Reads: line. The code, such as MK02, is the agent’s reference in the BearingBridge catalog, and its findings carry it too. The thirteen standard agents follow, with their default schedules. You can change a schedule once the agent is added.

The Standard section of the agent catalog, one card per agent with its code, what it reads, View details and Add to my agents
The Standard section of the agent catalog, one card per agent with its code, what it reads, View details and Add to my agents

Report agents

Both write a document every time they run. You read it on the agent’s card, and its main signals land in the findings feed as well. When a run covers a single brand, the report follows that brand’s guidelines. Unless you pick a model for them, both use the model your organization set for Ask Intelligence.

The Weekly operating review (DB13) runs every 168 hours by default. It reads Search rankings, Google Ads campaigns, Content pipeline, Social activity, Sales pipeline (CRM) and Platform spend, then writes up what changed and why it likely changed. It ends with three to five actions for next week. It runs for a brand that has at least one of these modules on: Google Search (SEO), Google Ads, Market Monitor or Content Plan.

The Executive report (MK16) runs every 720 hours, about once a month. It reads Search rankings, AI visibility (GEO), Google Ads campaigns, Content pipeline and Social activity, and writes the period’s narrative for leadership: a short executive summary and a section per channel with real numbers, closing on what happens next. It runs for a brand with Google Search (SEO), Google Ads, Content Plan or AI Search on.

Watchers

Watchers write findings, short alerts your team can act on. AI agents explains how findings work.

AgentWhat it doesBrand module neededDefault schedule
Search rankings watcher (MK02)Reports meaningful Google ranking moves: keywords entering or leaving page one, drops of three positions or more, newly ranked or lost keywords. Moves of one or two positions are ignored.Google Search (SEO)Every 20 hours
AI visibility watcher (MK03)Compares the brand’s visibility in AI assistants with its previous pass and reports gained or lost visibility, and competitors cited instead of the brand.AI SearchEvery 44 hours
Ads campaigns watcher (MK15)Flags anomalies in the live Google Ads campaigns: campaigns that stopped serving, status flips, budget changes, spend and cost-per-click jumps, with the real numbers.Google AdsEvery 20 hours
Social publishing watcher (MK10)Checks the publishing on every social network about once a day, network by network: what went out in the last 30 days, what is scheduled for the next 14, failed publications, scheduled posts that never went out, and drafts or posts awaiting validation that stalled. It judges each network the brand works on by itself, so a busy one never hides a silent one, and flags a network gone quiet and an empty schedule with the real dates. YouTube, uploaded by hand, is judged on what was marked published.Social Content or Social CalendarEvery 20 hours
LinkedIn publishing watcher (MK10)Checks the LinkedIn publishing about once a day: what went out in the last 30 days, what is scheduled for the next 14, failed publications, scheduled posts that never went out, and drafts or posts awaiting validation that stalled. It flags a quiet week and an empty schedule with the real dates.Social Content or Social CalendarEvery 20 hours
LinkedIn Ads watcher (MK09)Reads your own LinkedIn ad account, live, with your LinkedIn access, so it belongs in your personal agents. It compares the last 7 days with the 7 before and flags spend jumps, changes in cost per click and cost per result, campaigns that stopped delivering and budget pacing problems, with the figures and the currency.LinkedIn AdsEvery 20 hours
Market monitor (MK01)Re-reads the competitor search reports and ranking history, and reports competitor moves on the brand’s keywords, written for a marketing director.Market MonitorEvery 44 hours

One thing to know about the AI visibility watcher: each run makes one paid data call to fetch the citations the brand gained and lost per week. Every watcher runs on Claude Haiku 4.5 unless you pick another model.

Publishing agent

The Web editor (MK13) runs every 24 hours by default. It works on the sites of the Web module, each a GitHub repository. Each run starts the next brief in the queue of every active site. The brief is then researched, drafted, edited, rechecked, translated and illustrated, and published through a pull request that merges once the site’s own checks pass. It follows the rules kept in each site’s repository. Unless you pick a model for it, it uses your organization’s model for Web, then Claude Opus. It needs a Claude model to work.

Personal agent

The Verification codes agent (MAIL) runs every 24 hours by default, and belongs in your personal agents: it works on the mailboxes you connected on My Connections, which are yours alone. Services send a code at every sign-up, new address, sign-in check or password reset. Once the code has expired, the email is clutter, and this agent clears it away.

  • Outlook: each code email older than two hours is filed into a Verification codes folder of the same mailbox, created on the first run. A younger one may still be in use, so it waits for the next run.
  • Gmail: the app can’t move messages in a Gmail mailbox, so its code emails of the last 30 days are listed and stay where they are. Each one is listed once.

Each run lists the code emails it found, up to 40, with the service, the mailbox, the time and what the code was for: a sign-up, a sign-in check, a password reset or an email confirmation. It never shows a code, never uses one and never opens a link in the email. A sign-in check or a password reset is marked as a warning, because one nobody asked for means somebody tried that account, and the finding suggests changing that service’s password if nobody on your team signed in at that time.

The agent calls no AI model, so its runs cost nothing. On its Scope tab you untick the mailboxes it should leave alone, or hold it to Reads, in which case it lists the code emails and files nothing. An Outlook mailbox connected for reading only gets the same treatment until you connect it again from My Connections.

Library agent

The Library quality agent (AS01) runs once a week by default. It measures every picture and clip of the brand’s Assets Library, reading only the size of each file, and lists the ones too small to publish well: pictures under 1080 pixels on their short side and clips under 1080p, those held by a campaign first. Each finding gives the file’s size, the campaigns it is in, and how to fix it in Enhancement. It enhances nothing and spends nothing: it calls no AI model, so its runs cost nothing, and a run that finds the same files as the last one stays silent.

Listening agent

The Podcast and YouTube listening agent (VC01) runs every 24 hours by default. It checks the podcasts and YouTube channels your organization follows, listed in Voice of Customer with the words to listen for, and transcribes the episodes published since it was added. It never transcribes an episode twice: the first time it meets a show, the episodes already out are left aside. Each time your brands, your competitors or your chosen words are said, it reports the quote, the time and a link that opens the episode at that moment. A competitor named is a warning, everything else is information.

Its transcriptions are billed by the second of sound. A run listens to so many minutes at most (60 by default, set in Voice of Customer), and the agent stays within its monthly budget: 10 USD until you set another on its card. When the budget is spent, it waits for the next month. A podcast episode longer than the minutes left is heard in part; a video longer than a whole run is skipped.

New standard agents

From time to time the BearingBridge team adds a new standard agent to the app. It shows up in the catalog like the others and is opt-in too. Once added to your lists, its card carries the badge Standard · added by the platform.

Adding a standard agent

The whole catalog is open to every user, admins or not. Each card has two buttons.

ButtonWho sees itWhat happens
View detailsEveryoneOpens the agent as it ships, before you take it: what it does, Reads, Runs on (its model), Runs every (in hours), Web search (On or Off) and Instructions (what the agent reasons with).
Add to my agentsAnyone who can write: editors, brand managers and adminsOpens the agent form, filled with the agent’s standard setup. The agent joins your personal agents.

Viewers can read every agent in full, but can’t add one. Once an agent is among yours, the button turns into an In my agents link.

To add an agent:

  1. Click Add to my agents on its card, or at the foot of View details.
  2. A form opens, filled with the agent’s standard setup. It’s the same form used everywhere else in the Agents area: the Name on top, then four tabs. General holds What users are told it does and Instructions; Scope what the agent works on and its model; Schedule and budget Runs every (hours) and the two budgets; Notifications Email the results to and Email content.
  3. Change what you need. Leave a field empty to keep the standard version.
  4. Click Add the agent.

The agent is added switched on, under the Personal tab. That’s where you edit it, test it on real data and choose whether it reads several brands in one run. A watcher also gets Duplicate there, to make a variant of it; the two report agents don’t. Only the agents in your own list can be edited: the catalog stays as it ships. See organization and personal agents.

The catalog adds nothing to the organization. Administrators build the organization’s agents on the Organization tab, from its Add an agent list or from scratch, and every member then finds them on their Personal tab, under Your organization’s agents.

The data a standard agent reads is shown on its Scope tab but can’t be changed there. The rest of its scope can: switch a module or a connection off, hold it to reading, or change the model. To choose the data yourself, build a custom agent.

Tip: Not sure an agent is worth it? Add it to your own agents first and test it on a real brand. The test shows what it would report and what it costs, and posts nothing to the brand’s feed.

Tailor-made agents

The rest of the page is a reference list. Skim the headings for your area.

Tailor-made agents aren’t platform features. Each one is built for one organization, on its own data and rules. Some already run for other organizations; the rest get built when someone asks. That’s also how internal IT support is handled, along with every Service Desk agent beyond the self-service assistant.

In the catalog they sit in the Tailor made section, each card marked Tailor made only. Use the category chips above the grid to show one area at a time, or All. Each chip shows how many agents it holds. Each card has a Contact us to build it link that opens the contact form with the request filled in. The app shows no price for these agents. The contact form is where the conversation starts.

The Tailor made section of the catalog, with one category chip per business area and a Contact us to build it link on each card
The Tailor made section of the catalog, with one category chip per business area and a Contact us to build it link on each card

Sales, finance and operations

AgentAreaWhat it does
Proposal draft agentSalesDrafts a client-ready quote or proposal from a deal’s CRM context: client, amounts, next steps and activity history, in your templates and tone.
Quotation agentSales and pricingAssembles quotations from client context and written pricing logic; sales checks and sends.
Sales forecast agentSales operations and financeScenario forecasts as ranges with logged assumptions, owned and decided by people.
Month-end close agentGroup financeCoding, matching and commentary drafting for the monthly close, with the ledger itself left untouched.
Supplier document matching agentProcurement and supply chainExtracts supplier documents and three-way matches them at confirmation, with an exception queue for people.
Executive dashboard agentExecutive reportingOne dashboard over every system, with agreed metric definitions, that leadership can question in plain language.

Marketing and strategy

The Market monitoring agent (Strategy) reads market sources across languages and writes weekly briefings backed by dated primary sources. The Campaign data agent (Ad operations and analytics) normalizes campaign data across ad platforms against a metric dictionary and shows where sources disagree. The Ads factory agent (Demand generation) drafts, tests and adjusts campaigns across ad platforms from one workspace. The Content factory agent (Marketing) produces on-brand assets at scale from brand context packs, with a hard legal gate before anything ships.

Service Desk

These run on your own channels, helpdesk and order systems.

AgentWhat it does
Customer inquiries agentTriages, retrieves and drafts answers to routine customer inquiries, under a conservative escalation rule.
Omnichannel inbox agentBrings email, chat, voice, messaging apps and social into one queue, with the same customer recognized across all of them.
Response drafting agentDrafts the reply from your own history and tone of voice; the person handling the case decides what goes out.
Response library agentKeeps macros, templates and snippets in one place and offers the right one inside the conversation.
Order and account lookup agentPuts the customer’s orders, account and past contacts in front of the person answering.
Warranty and returns agentCarries a warranty, return or claim from request to resolution, with the exceptions queued for a person.
Service SLA agentWatches targets and timers per conversation and escalates on your own paths before a promise is broken.
CSAT agentAsks after contact, collects the answers and shows what moved satisfaction, not just the average.
Service QA agentReviews interactions against your quality grid and shows each score with the passage it came from.
Coaching agentTurns quality reviews into coaching for each person, with examples from their own conversations.
Customer health agentBuilds an account-level service picture: volume, sentiment, open issues and where the relationship stands.
Multilingual service agentHandles inbound and outbound in the customer’s own language, written natively rather than translated.
Voice of the customer agentReads what customers say across every channel and reports the themes with the quotes behind them.
Proactive outbound agentContacts the customer before they complain, on the signals you decide are worth a message.
Complaint handling agentTracks serious cases from first contact to closure, with deadlines, owners and the record a regulator would ask for.
Service report agentReports volume, resolution, cost and quality in one place, on definitions everyone has agreed.

The self-service assistant itself ships in the app. See the Service Desk.

IT support

Sixteen agents for internal IT, built against your own tools and rules.

AgentWhat it does
Help desk agentTakes in employee tickets, sorts them into the right queue and assigns an owner, with the whole thread kept on the ticket.
IT self-service agentAnswers employees from your own runbooks and documentation, and opens a ticket only when it can’t resolve the request.
IT knowledge base agentTurns resolved tickets into articles, runbooks and procedures, and keeps them current as your systems change.
Ticket triage agentReads each incoming request and sets its category, priority and owner against your own routing rules.
IT remediation agentRuns a bounded set of fixes you have approved in writing, and hands anything outside them to a person.
Access request agentCarries an access request from ask to approval, provisioning and revocation, with every step recorded.
Joiner, mover, leaver agentRuns the onboarding and offboarding runbook for each arrival, move and departure, and checks nothing was left open.
IT asset agentKeeps devices, their owners and their lifecycle in one inventory, and flags what has gone missing or out of date.
Software license agentTracks seats, cost and real usage across your tools, and shows what can be given back at renewal.
Infrastructure monitoring agentWatches servers, services and availability, and reports what changed in plain language instead of raw alerts.
Incident management agentDeclares an incident, keeps the timeline and the people on it coordinated, and records how it was resolved.
Patch and vulnerability agentSays what is exposed, what is already fixed and what is still waiting, against your own risk rules.
Change management agentTakes a change request through risk assessment, approval and record, with the evidence attached to it.
On-call agentRuns the rotation, the escalation ladder and the handover note, so nothing is dropped between shifts.
IT SLA agentTracks targets, breaches and trends per service, and warns before a target is missed rather than after.
Postmortem agentDrafts the postmortem from the incident record: what happened, why, and the actions somebody owns.

Good to know

  • Adding an agent costs nothing. Its runs are billed as they happen and each one shows its cost. A watcher run that finds nothing changed is free. Budgets per agent are covered in organization and personal agents.
  • The catalog always lists every standard agent. Once you remove one from your Personal tab, its Add to my agents button comes back and you can add it again at any time.
  • An added agent only runs on brands that have its module switched on. If none does, its card on the Organization tab says so.
  • If the BearingBridge team removes a standard agent from the app, it disappears from the catalog and stops running everywhere.

Roles are explained in roles and rights, and prices in credits and costs.