AI skills
Teach your AI features how your team works by writing reusable playbooks Remie, research, and reply drafts follow.
AI skills are short playbooks you write once and your AI features reuse everywhere. A skill captures how your team actually works: how you qualify a lead, handle a common objection, or run a first call. Remie, contact research, and reply drafts all read your enabled skills, so their output follows your process instead of a generic default.
Skills live under Settings, AI skills and require the Manage settings permission.
How skills are used
Every skill has a short name, a one-line description, and a playbook (the full instructions, written in plain markdown).
When an AI feature runs, it is shown the name and description of each enabled skill. When one is relevant to what it is doing, it loads the full playbook and follows it. This keeps prompts short while still giving the AI access to everything you have written down. A skill you switch off is hidden from the AI entirely.
Writing a good skill
- Keep the name short and specific, like "Objection handling" or "Discovery call".
- Make the description a single line about when to use it, so the AI knows when it applies.
- In the playbook, write plain, direct instructions. Bullet points and short steps work well. Say what to do and what to avoid.
Example: objection handling
Name: Objection handling
Description: How to respond when a prospect pushes back on price or timing.
When a prospect says it is too expensive:
- Acknowledge it directly, do not get defensive.
- Ask what they are comparing the price to.
- Reframe around the cost of the problem staying unsolved, with a specific number if you have one.
- Offer a smaller first step, never a discount as the first move.
When a prospect says "not right now":
- Ask what would need to be true for it to be the right time.
- Offer to send one useful thing and check back, do not push for a meeting.
- Keep it short and low pressure.Example: discovery call booking
Name: Discovery call booking
Description: How to ask for and frame a first call.
- Only ask for a call once the prospect has shown a real signal of interest.
- Frame it as a short, specific conversation, not a demo. "15 minutes to see if this is even a fit."
- Always offer an easy out in the same message ("or if this isn't a priority, no worries").
- Never stack asks or send more than one call link.
- Match our voice: direct, warm, lowercase openers are fine.Playbooks for the inbox agent
The inbox agent works differently: you pick one skill as its playbook under Settings, Workspace, Inbox agent, and the agent reads the whole of it on every reply instead of loading it on demand. Auto-send needs one, and only an enabled skill with content can be chosen.
Split the playbook into sections with ## headings. Each section becomes something the agent can cite: every question a prospect asks is answered from a named section, from the facts the agent is given about you and the contact, or with a plain "I can't confirm that here". A reply that cites a section that does not exist is never sent.
We sell Zexr, a cold email platform for small sales teams.
## Pricing
Plans start at $49 per seat a month, billed annually. Every plan includes warmup.
## Onboarding
Most teams send their first campaign within a day. Setup steps: https://zexr.app/docs/start
## Objections: too expensive
Compare it to the cost of one SDR. Offer a smaller first step, never a discount first.- Text before the first heading is shared context; it is read but cannot be cited.
- A level-three heading (
###) stays inside its section. - A section id is its heading in lowercase with dashes (
## Objections: too expensiveisobjections-too-expensive), and a repeated heading gets-2,-3. - Write the facts you want quoted, in full: prices, terms, links. The agent states nothing about your product that is not in the playbook.
- Put any link the agent may share in the playbook, and its host on the agent's allowed links list.
- The playbook is never shown or mentioned to a prospect. A skill holds up to 192 KB, and the agent reads all of it.
Where skills apply
- Remie: follows relevant skills when it drafts, plans, or answers.
- Contact research: uses skills to judge what is worth surfacing about a contact.
- Reply drafts: writes replies that follow your playbooks.
- Inbox agent: answers replies from the one skill chosen as its playbook.
Because skills are shared across your workspace, one person can capture how the team works and everyone's AI output improves at once.
Automatic inbox tagging
Label and score inbound mail as it arrives, keep automated mail out of the inbox, and let a confident verdict hold, stop, follow up or suppress. Optional, off by default.
Connect MCP tools
Give Remie extra tools by connecting an external MCP server, reviewing what it exposes, and turning it on.