DeskFerry is a no-code AI agent platform whose core consists of two parts: "AI employees" that perform specific work and an AI assistant (Assist) for quick questions. Without writing any code, users can describe in natural language the work content for each AI employee, the apps available to it, and the actions that require user confirmation, enabling it to complete tasks autonomously.
Purpose and main capabilities:
- AI employees: Each AI employee can take on a complex function, such as customer support, inbound lead qualification, invoice collection, content marketing, and more. They read, write, and perform actions across more than 1,500 apps, including Gmail, Slack, HubSpot, Salesforce, Zendesk, Notion, Stripe, Google Sheets, and Jira.
- Task board: The work of all AI employees is centralized on one board, divided into Backlog, Needs You, In Flight, and Done. Users can plan tasks and assign them to AI employees, and cards move automatically as work progresses, with no need to refresh or manually track them.
- Approval mechanism: Any action that sends, spends, or modifies records pauses and waits for the user to approve or reject it in Desk, so users always remain in control of key decisions.
- Scheduling and triggers: Supports scheduled runs such as hourly, daily, and weekly, and can also be triggered by events in other apps.
- Knowledge, memory, and logs: Documents, databases, and URLs can serve as task context; preferences and past conversations are retained; every run records what was done, which apps were touched, and where it paused, avoiding silent failures.
- Model selection: Built-in models such as OpenAI, Anthropic, Google, and xAI can be switched per task, with no need to manage API keys.
- Custom skills: Skills such as brand tone and report formats can be added and shared with the team.
- Assist: A general-purpose AI assistant that can handle many simple tasks, supports chat and tasks such as scheduling, and comes with its own Skills and Notes.
- Multiple entry points: AI employees can be triggered through website widgets, API calls, Slack messages, or email. In Slack, @DeskFerry and say the task name to launch the corresponding AI employee.
How to use it: Users write down tasks as if assigning them to a colleague and select the corresponding AI employee; the AI employee takes over and executes them, and the board updates in real time; actions that need to send, spend, or modify records will be handled in Desk. Waiting for confirmation; once completed, the task moves to Done. You can select preset sales, support, or operations AI employees from the Marketplace, or describe a new employee in natural language; each AI employee learns the user's preferences over time.
Use cases:
- Sales: inbound lead scoring, data enrichment, routing hot leads to Slack.
- Customer support: instantly handle tier-one queries, escalate the rest with full context.
- Finance/operations: track invoices from issuance to payment, send reminders, flag accounts that need human intervention; month-end close reminders, etc.
- Content marketing: draft content in the user's tone, handle SEO structure, and publish to WordPress, Ghost, or Webflow after approval.
- Data analysis: ask questions in natural language, and the AI employee queries across the CRM, spreadsheets, and billing systems and answers directly in chat.
Limitations and notes:
- No free plan; a one-time 7-day free trial is offered. Paid plans are Growth $49/month, Pro $99/month, and Scale $299/month. Paid plans renew automatically until canceled; see the official website for the refund policy. All plans include an unlimited number of AI employees; the difference is the amount of work that can be completed each month.
- Data is processed in a secure isolated environment with enterprise-grade encryption, and data is isolated within the user's own workspace.
- The materials mention that supported models include OpenAI GPT, Anthropic Claude, Gemini, Grok, and open-source models; the specific choice depends on the use case and privacy needs.
- Other features, pricing details, or platform information not explicitly stated in the materials are not yet provided.
