Customer relationship management is where a business keeps its people — contacts, accounts, deals, tickets, conversations. The 2nth CRM skills turn that graph into an operational substrate for AI: one GTM role at a time, one bounded tool set at a time, every mutation a human approves.
HubSpot is the single-platform customer graph most mid-market businesses end up on. Salesforce is the enterprise default with its own AI framing in Agentforce and Einstein. The role model and operational pattern are shared — the API shape is not.
The CRM that grew into a customer platform — sales, marketing, service, CMS, and operations on one object graph. v3 REST API, Private Apps for internal integrations, Associations v4 for the graph edges, Search API for anything that needs filters. The canonical skill maps six GTM roles to bounded MCP tool sets.
The enterprise CRM default. Real SObject model, SOQL queries, Apex hooks, and the whole Agentforce + Einstein AI layer baked into the platform. Slotted for real content once the auth patterns, SOQL templates, and Agentforce-vs-2nth-role mapping are ready.
Both tools solve the same problem, differently. The comparison matters the first time someone has to choose — and matters even more when a business runs both and needs them to agree on who a customer is.
| Skill | API style | Auth | Strongest for | Native AI surface |
|---|---|---|---|---|
| HubSpot | REST v3 + Associations v4 + Search API | Private App token (internal) / OAuth (marketplace) | Mid-market GTM on one platform, five hubs, unified object graph | Breeze |
| Salesforce | REST + SOQL + Bulk API 2.0 + Streaming | OAuth 2.0 (Connected App) / JWT Bearer | Enterprise CRM with deep customisation, Apex, multi-org complexity | Agentforce + Einstein |
CRM is not ERP, but the 2nth shape is close. Read the customer graph, enrich it with AI reasoning, write back only what a human has approved. The value is never in autonomous mutations on pipeline and money — it's in diagnosis, drafting, and hygiene.
| Step | What it does | Canonical principle |
|---|---|---|
| 1 · Authenticate | Private App token, JWT, or OAuth — scoped narrowly. One token per role, never one god-token. | Scopes are the real security boundary |
| 2 · Discover | List properties, pipelines, custom objects. Internal names differ from UI labels — always fetch first. | Every portal is bespoke · never assume schema |
| 3 · Query | Use Search APIs (HubSpot) or SOQL (Salesforce). Filter server-side. Cap-aware cursoring. | Pagination caps bite hard · split by time window |
| 4 · Enrich | AI reasoning on the returned objects — scoring, summarising, drafting, classifying. | The AI reads · the human decides · the API writes |
| 5 · Write | Human-approved mutations only. Log every change. Associations v4, not v3. | Pipeline and money need human approval |
The role model is the thing 2nth contributes on top of the raw CRM APIs. Every leaf in this sub-domain gives each role its own agent with its own narrowly-scoped tools — the human stays accountable, the AI handles the volume.
| Role | The human decides | The AI enables |
|---|---|---|
| Revenue Leader | Forecast, territory, quota, GTM strategy | Pipeline health, win-rate drift, stalled-deal triage |
| Sales Rep | Deal relationships, call priorities, closing moves | Next-best action, draft follow-ups, meeting summaries |
| Marketing Manager | Campaign strategy, positioning, budget | Segments, subject-line A/B, attribution reports |
| Service / CS Lead | Escalations, voice of customer, NPS | Ticket routing, sentiment flags, draft replies |
| RevOps Engineer | Object model, process design, pipeline stages | Property audits, workflow review, association mapping |
| Executive / Owner | GTM priorities, resource allocation, board narrative | Revenue dashboards, campaign ROI, churn signals |
CRM sits in the middle of the GTM layer. It requires at least one skill from tech for the runtime, borrows customer data from biz/erp, and feeds pipeline numbers up into fin/reporting when that branch goes live.