👨‍🔬How to Launch Calk AI Internally

1

Find Your First Use Case (Keep It Small)

The best AI use cases are:

• Repetitive

• Time-consuming

• Based on internal knowledge

👀 Start here:

• New hire onboarding

• Answering common product questions

• Internal help desk (“Where’s the doc for X?”)

• Sales enablement bot (CRM + battlecards)

📎 Bonus: Check our Use case page ⇒ Here 🎸

2

Choose the Right Sources to Connect

Don’t overdo it at first — start with 1 or 2 tools like:

✅ Notion → for wikis, SOPs, meeting notes ✅ Slack → for tribal knowledge in channels ✅ Hubspot or Intercom → for sales/support agents ✅ Google Drive → for project docs or specs

🎯 Pro tip: The more focused the data, the better the answers.

3

Create Your First Agent (in Minutes)

You’ll:

• Give it a name (e.g. “SupportBot”)

• Choose a model (start with Claude or GPT-4o)

• Add instructions (e.g. “You help answer product questions using our docs.”)

• Link 1–2 sources of content

That’s it. Ask your first question and test it live 🎉

4

Test Internally Before You Invite Everyone

Start solo or with 1–2 teammates. Try these:

• “What’s our onboarding checklist?”

• “How do we handle refunds?”

• “What did we last say to client X?”

Adjust your sources or instructions based on what it says.

5

Invite Your Team + Track Use

Once the first agent works, invite others:

• Support team → connect Intercom

• Sales → connect CRM

• Ops → connect wikis or SOPs

Calk will feel like a second brain, not a project.

💬 Where to Learn, Ask & Share

🔗 Join our Slack community

Ask us anything, share your agents, or see what others are building.

📺 Watch our YouTube tutorials

Get inspired by real-world use cases and tips from the team. Here

🧰 Access our Notion toolbox

Copy-paste-ready prompts, agent templates, and best practices. Here

---

✅ TL;DR

1. Start with 1 use case per business unit

2. Connect only the tools that matter

3. Create a focused agent

4. Test it, refine it, share it

5. Get your team involved

➡ find our use cases here 🎸

---

🎬 Ready for Part 3? We’ll go deeper into building better agents, customizing behavior, and keeping knowledge synced effortlessly.