What Swan does
- Research and enrichment. Swan looks up companies and contacts, pulls in firmographic and contact data from its data providers, and fills in gaps on records you already have.
- Prospecting. Swan finds new companies and contacts that match your target market, using buyer-intent signals and sourcing tools.
- Outreach sequences. Swan writes and runs multi-step email and LinkedIn sequences — drafting messages, sending them through your connected senders, and reacting to replies.
- CRM work. Swan reads and writes CRM data through its integrations catalog — most deeply HubSpot, plus dozens more (Salesforce, Slack, Attio, Google Analytics, and others) — and any custom MCP server your org connects.
- Triggers (automations). You can set up standing automations — instructions plus a set of toolkits plus a source (a schedule, a webhook, a CRM event, a website visitor signal, and more) — so Swan reacts to things happening in your systems without you asking each time.
- Desk tasks and Slack. When Swan needs a human decision, it creates a task on the Desk or posts a Slack notification instead of guessing.
- Analytics. Swan tracks credit usage, execution history, and per-contact outreach status.
- Skills and templates. Skills are reusable instructions you or the Template Library provide that Swan follows when doing a specific kind of work; templates are ready-made plays you can install and adapt.
- Presentations and code execution. Swan can build presentations and run code as part of a task — transforming a dataset, generating a file, or calling an external API directly with a stored secret when there’s no ready-made integration (see direct API access).
What Swan does not do
Swan is deliberately scoped. It is not a general-purpose product it wasn’t built for:- No SMS. Swan sends email and LinkedIn messages, not SMS.
- Not a website chat widget. Swan doesn’t sit on your website answering your visitors’ questions. It’s an internal GTM agent your team talks to, not a customer-facing chatbot. (Swan does support tracking and identifying anonymous website visitors as a data source — see the website visitors trigger — but that’s a signal for your team, not a chat surface for your visitors.)
- No custom dashboard builder. You can’t build your own charts or dashboards inside Swan. Use the built-in analytics surfaces, or send data to your CRM/BI tool for custom reporting.
- No white-label option. There’s no custom branding or reseller product.
- No public REST API. Swan doesn’t offer a REST API or bulk-export endpoints. The programmatic surfaces are Swan’s own MCP server (point your agent at Swan and hand it work) and webhook triggers (inbound). If you need data flowing out of Swan, the supported paths are your CRM integration, or an automation whose instructions have the agent call your systems’ APIs with a stored secret — see direct API access.
- No scraping behind logins with your credentials. Swan doesn’t accept your username and password to log into a website and scrape it on your behalf. Its web research tools work against public data; official OAuth integrations are the supported way to pull data from platforms you use.
- Not reliable for precise math. Swan is built on large language models, which are not a good fit for exact calculations or deterministic logic. For anything that needs to be precisely correct, use a code execution step, an external tool, or your CRM’s own workflow logic rather than trusting the model’s arithmetic in prose.