brand/agent/control_
Projects • CRM • Marketing • Memory • Approvals • MCP

Give your AI agent a real place to work.

Sally runs your brand's projects, CRM, marketing, and operational memory — and connects to the AI agent you already use over MCP. Your agent does the work. You approve what matters.

No bundled bot, no lock-in: bring Claude or any MCP-capable agent. Sally gives it real tools, real context, and hard approval boundaries.

7-day free trial. No credit card required.

sally/mcp
> tools/call project.runtime_context
← project summary · open tasks · blockers · memory · safe actions
> tools/call task.create { "title": "Draft June client newsletter" }
← task created · status: In Progress
> tools/call marketing.message.propose_send
← approval requested: PUBLISHING · nothing sent
your agent
Reads context, drafts, updates tasks and CRM, files handoffs
approval gate
Human approval required before anything risky happens

Bring your own agent

Sally ships no bot. It ships the workplace.

You already chose an agent. Sally exposes its full product surface — projects, tasks, CRM, marketing, memory — as MCP tools your agent can call, under your permissions.

Connect in minutes

Create an MCP key in Sally or connect over OAuth, point your MCP client at your Sally endpoint, and your agent can list and call real Sally tools.

Same rules as humans

Agents work under the same account, workspace, and project permissions as your team. No side doors, no shadow state.

Revocable and auditable

Keys can be scoped and revoked, agent sessions are visible, and every action lands in the activity and audit trail.

How Sally works

Not another blank prompt. A context-backed operating loop.

1. Work happens

Projects move, tasks close, meetings get captured, customers ask questions, campaigns send. It all lives in Sally per brand.

2. Sally spots the next action

The opportunity engine surfaces what the activity supports: a newsletter draft, a client status report, a follow-up, a briefing.

3. Your agent does the work

Connected over MCP, your agent reads runtime context, drafts, updates tasks and CRM records, and files structured handoffs.

4. You approve what matters

Sending, publishing, credentials, customer data, and destructive actions raise typed approval requests in Control Center.

5. Sally remembers

Decisions, lessons, brand voice, and campaign learnings become durable memory that improves the next piece of work.

One platform per brand

Everything a brand runs on, in one operable system.

Each brand in Sally gets the full surface. Every part of it is available to humans in the UI and to agents over MCP.

Projects & delivery

  • Boards with custom statuses and priorities
  • Rich task descriptions, checklists, comments, and mentions
  • Task and project dependencies, timesheets, and client links

CRM

  • Organizations, people, deals, and activities
  • Follow-up reminders so conversations do not go cold
  • Relationship context shared across your brands

Marketing

  • Contacts with custom fields and GDPR consent
  • Visual email builder and graph-based campaigns
  • Embeddable forms, content gates, and delivery feedback

Memory & opportunities

  • Durable memory: facts, decisions, lessons, brand voice
  • Meeting notes and transcripts become CRM records and tasks
  • Opportunity wedges turn activity into concrete next actions

Human control

Agents propose. Humans decide. Sally enforces it.

Control Center collects proposals, approvals, blockers, decisions, and risks in one review surface — so letting agents work never means losing the plot.

Typed approvals

Live deploys, credentials, payment data, customer data, client decisions, destructive actions, and publishing each require explicit human approval.

Trust policies

Decide per agent which safe actions run automatically and which always wait for review — from approve-everything to hands-off.

Typed blockers

Agents raise structured blockers — missing access, failing tests, ambiguity, client decisions — instead of guessing.

Security baseline

SSO, two-factor and passkeys, session and key policies, and an audit log across sensitive actions.

Pricing

Every plan includes everything. Pay per brand, not per seat.

Solo
€250per month

1 brand. The full platform for a single brand or business.

Team
€500per month

3 brands. For teams and agencies running several brands.

Growth
€1,000per month

7 brands. For multi-brand operations at scale.

next/action

Put your agent to work inside your business — safely.

Create your account, set up your first brand, connect your agent over MCP, and keep every risky action behind an approval you control.

7 days free. No credit card required.