Agent library
Every Unbound system is assembled from a real, working roster. Not a concept deck. 43+ agent roles across nine categories, each one already running inside a live operation before it ever reaches yours.
Most of this library is Rose Wild's own agent system, built and run in her day-to-day operation before it was templated for Unbound clients. Every role below reflects what an agent actually does in production, not what it could theoretically do.
A few roles started as thinner scaffolds rather than fully proven builds. Those get shaped to your business during the build process, the same as everything else.
Start here
2 roles
Best for: Any business that wants one place holding the whole operation.
Central intelligence
Holds the whole operation, remembers what happened, and hands off every piece of work to the right specialist agent instead of doing it all itself.
Accountability
Works alongside the Commander to turn decisions into action, follow through on what got assigned, and hold every other agent to it.
4 roles
Best for: Any system running agents that touch real data, money, or credentials.
Security agent
Watches for threats, suspicious activity, and exposed credentials, and hardens the system before something goes wrong.
Nightly audit
Checks the Commander's own files and settings every night for anything that changed without you knowing.
Scope guard
Keeps an eye on an outbound pipeline for scope creep, exposed client data, or configuration drift.
Trust ladder
Reviews how every agent actually performed and promotes or pulls back its autonomy based on results, not a guess.
6 roles
Best for: Brands publishing content on a real cadence across social and email.
Strategy
Plans and oversees content and outreach strategy across every brand you run. Never posts or spends on its own.
Copy
Writes social posts, captions, and long-form copy in your exact voice.
Analytics
Reads your social performance data and reports what is actually working in a weekly briefing.
Distribution
Takes approved content and schedules it out across platforms. Never writes copy, only ships it.
Lifecycle email
Runs welcome sequences, broadcasts, and list cleanup, using copy your content writer already approved.
Asset finishing
Turns raw photos and footage into finished, ready-to-post graphics and video.
6 roles
Best for: The team that keeps your agent system built, maintained, and growing.
Development
Designs, builds, and maintains the agents inside a brand's system, scoped to whichever part of the business it owns.
Dashboards
Builds dashboards and internal tools for the Commander system itself.
Client-facing design
Builds decks, landing pages, and visual assets for a specific brand, locked to that brand only.
Read-only
Reads the codebase and documentation to gather what a build needs. Never writes code, never touches a credential.
Implementation
Writes the code and documentation for a change, one clean piece at a time.
Quality gate
Tests every change against real data before it ships. A gate, never a rubber stamp.
10 roles
Best for: Businesses running outbound sales or lead generation at real volume.
Diagnosis
Looks at where you are still stuck doing the work by hand and proposes what to automate next.
Outbound orchestration
Oversees an outbound sales pipeline end to end and flags what is broken before it costs you leads.
Research
Researches prospects and builds the profile outreach gets written from. Never sources an email address itself.
Cold email
Writes cold emails and follow-ups from a researched profile, following the approved playbook every time.
QA
Reviews outreach copy and research before it goes out. Never ships anything itself.
Delivery
Packages a finished outreach batch into one deliverable and writes the summary.
Intake
Turns a new form submission into a usable profile for the next step in the pipeline.
Free trial
Turns a lead's intake into a few sample emails and a short business reflection, ready to send as a taste of the work.
Pre-campaign
Validates a target market or client fit before a new outbound campaign launches.
Platform monitoring
Watches lead platforms and drafts responses and quotes for your approval.
8 roles
Best for: Anyone drowning in email, calendar, and CRM upkeep.
Email agent
Manages your inboxes: reading, sorting, flagging, and drafting replies in your voice.
Calendar agent
Manages your schedule across every calendar you run and flags conflicts before they happen.
Operations
Works inside your CRM: reads records, posts updates, creates tasks, and manages mentions.
Capture
Turns handwritten or dictated notes into CRM records automatically.
Data quality
Checks CRM records for missing fields and fixes what it can safely fix on its own.
Morning briefing
Delivers a daily rundown of what happened and what's next, in plain language, first thing in the morning.
Tech support
Handles day-to-day Mac and phone housekeeping: settings, contacts, app management.
Memory
Captures what happened in a work session and writes it into the system's permanent memory.
2 roles
Best for: Businesses that need real research before making a call, not a guess.
R&D
Handles market analysis, competitive research, and pricing intelligence on demand.
Capability watch
Watches for new AI capabilities and tells you which ones would actually improve your system, not just what's new.
2 roles
Best for: Systems complex enough that something breaks quietly if nobody's watching.
Infrastructure
Runs on a schedule, catches infrastructure problems, and fixes what's safe to fix without asking.
Queue management
Works through a queue of known issues in priority order and routes each one to the right builder.
3 roles
Best for: Agencies and freelancers running fully separated work for their own clients.
Content
Writes a client's social content in the client's exact voice, fully separated from your own brand work.
Distribution
Turns a client's approved content into a ready-to-publish schedule.
Analytics
Tracks a client's social performance and feeds it back into their content plan.
Agents from this library ship as part of an Unbound package, built and scoped to your operation, not handed over as a raw template. A Foundation build starts with one Commander Agent. Larger builds add the specialists your operation actually needs, drawn from the categories above.