> ## Documentation Index > Fetch the complete documentation index at: https://docs.apollospace.ai/llms.txt > Use this file to discover all available pages before exploring further. # People & CS > Onboarding, support, and relationship continuity that don't depend on one person being online — with the sensitive line always staying human. It's 8am on a Monday. A brand-new customer messaged on WhatsApp — "hey, I started yesterday and got stuck on step two"; a colleague joined the team today and doesn't know where anything lives; and two people who asked for support on Friday still have no answer because whoever handled it is off. All of these are relationships cooling off — not from bad intent, but because they hinge on one specific person being on call. **What leaves your plate:** first-line support, the onboarding walkthrough, the follow-up nobody remembers to send, and logging every conversation in the right place all start happening on their own — with context, within budget, and with the sensitive part always waiting on a human. ## How it works, in one line People and CS is, at bottom, **memory that can't die with one person.** The agent doesn't replace whoever owns the relationship — it makes sure the context, the continuity, and the first response exist even when that person isn't around. ```mermaid theme={null} flowchart LR A[Trigger
new person, question,
check-in date] --> B[Agent pulls context
from Company Brain + CRM] B --> C[Decides + uses tools
Chats, CRM, Routines, Docs] C --> D{Sensitive?
comp, contract,
personal} D -->|No| E[Answers / logs / schedules] D -->|Yes| F[Proposes + calls a human
with full context] E --> G[Becomes memory
the next interaction already knows] F --> G ``` ## Concrete flows ### 1. Run the onboarding — for a customer or a teammate **Trigger:** a new person arrives — a customer who just signed, or a colleague on day one. [Athena](/en/agents/athena), the organization's operator, pulls the onboarding script for that situation from the [Company Brain](/en/features/brain) — the steps, the documents, who owns what. She opens a [chat](/en/features/chats) and walks the person through it **one step at a time**: "registration done? then X is next." As each step clears, she creates the matching [tasks](/en/features/boards) on the onboarding board, attaches the right [document](/en/features/documentos), and checks off what's complete. If the person gets stuck on something the Brain doesn't cover, Athena **doesn't make it up** — she logs the question and escalates to the human owner with the history of what was already tried. When it's done, the whole onboarding became [memory](/en/features/brain): next time the script is sharper. The same pattern serves both audiences. **Customer** onboarding = activate whoever bought so they see value fast. **Teammate** onboarding = get the person out of "I don't know where anything is." The difference is the script in the Brain, not the mechanics. ### 2. First-line support — answers from the Brain, escalates the rest **Trigger:** a question lands by [chat](/en/features/chats) or [WhatsApp](/en/integrations/whatsapp). The agent looks for the answer in the [Company Brain](/en/features/brain) — FAQ, policies, product docs the org already ingested. When the answer **is there**, it replies on the spot, in the right language, and logs the interaction as an [activity in the CRM](/en/features/crm). When it **isn't** — or when the case is clearly sensitive — it stops, doesn't guess, and **hands off to a human** with everything chewed: who the person is, what they asked, what it tried, and why it stalled. ```mermaid theme={null} flowchart TD A[Question arrives] --> B[Agent searches the Brain] B --> C{Found a
trustworthy answer?} C -->|Yes| D[Replies + logs in CRM] C -->|No / sensitive| E[Human handoff
with full context] D --> F[Becomes memory] E --> F ``` The win isn't "a bot that answers everything." It's that **the first response never lags again because someone was busy** — and what needs a human reaches a human already in context, not as a "remind me what happened again?" ### 3. Check-in and NPS — the follow-up nobody remembers to send **Trigger:** a [routine](/en/features/rotinas) the agent schedules for itself — 7 days after onboarding, 30 days later, end of every month for the active book. On fire, the agent pulls from the [CRM](/en/features/crm) who's at the right point in the cycle, drafts a check-in message or a short satisfaction question, and sends it on each customer's channel ([email or WhatsApp](/en/features/outbound)) — with **human approval before the send** if you want to keep a hand on outbound. Replies flow back into the CRM as activity; anyone who scored low or flagged dissatisfaction becomes a prioritized [task](/en/features/boards) and a [proactive notification](/en/features/rotinas) in the account owner's bell. **Why this matters:** relationship follow-up is the easiest work to defer and the costliest to forget. A routine has no hectic Monday — it sends the check-in on the right day, every time, and turns each reply into an actionable signal instead of a spreadsheet nobody reads. ### 4. Keep customer context alive — and the handoff that doesn't restart from zero **Trigger:** any interaction ends — a chat, a logged call, an email exchange. The agent updates the [CRM](/en/features/crm) right after: logs the activity, adjusts the stage if the relationship moved phase, attaches a summary of what was agreed, and stores the learning in the [Company Brain](/en/features/brain). The result: the customer's history is **always current**, not dependent on someone remembering to write it up later. And when a case gets hard — a delicate negotiation, a serious complaint, something that needs human judgment — the handoff carries that whole context. Whoever takes over doesn't ask "who's this customer again?": they open the card and see the full timeline, what was promised, the tone of the last conversation. **Continuity stops living inside one person's head.** ## What stays human The agent **proposes**, never decides alone, whenever the subject is: * **Compensation** — salary, bonus, raise, any money conversation with a colleague. * **Hiring and exits** — onboarding, offboarding, promotion, any decision about people. * **A contractual promise to a customer** — deadline, scope, commercial term, any commitment that becomes an obligation. * **Customer money** — refund, discount, charge, invoice adjustment. * **A sensitive personal matter** — health, conflict, serious complaint, anything that calls for human empathy and judgment. * **Heavy external, legal, or regulatory communication**, governance decisions, and destructive actions. In those cases the agent prepares everything — the draft, the context, the recommendation — and **waits for a person to approve.** Every action lands in the auditable trail, with author and justification. ## Where to start Upload your onboarding material and the support answers you repeat every week into the [Company Brain](/en/features/brain). That's where the agent pulls its first response from — without it, there's nothing to consult. Connect [WhatsApp](/en/integrations/whatsapp) or [chats](/en/features/chats) and let the agent answer what's in the Brain, escalating the rest. Start with human approval before each external reply and loosen it as it earns trust. Create a follow-up or NPS [routine](/en/features/rotinas) for your active book. One agent, one cadence, one channel — and autonomy grows like a ratchet, not a leap. ## Next steps Where support and onboarding happen — the org's operator one keystroke away. Where customer context stays alive — leads, activities, and history that don't depend on human memory. The cron behind check-ins and follow-ups — the agent remembers for you. The same support and post-sale pattern applied to a high-volume sector.