WisebotAI
Use cases & roadmap

Dogs use case - first agent guide

A guided, screenshot-backed walkthrough to create your first WisebotAI agent and test it end-to-end.

Dogs use case: create your first agent

Use this as a guided first run for a "Dogs assistant" that answers adoption, care, vaccination, and grooming questions.

What you will build

  • One agent: Dogs Assistant
  • One knowledge folder for dog-care content
  • One test channel for validation
  • One baseline workflow/tool setup for real answers

Step 1: Create the agent

  1. Open Dashboard -> Agents -> Create agent.
  2. Name it Dogs Assistant.
  3. Set role prompt:
    • "You are a helpful dog-care assistant. Be concise, safe, and practical."
  4. Save and continue to configuration.

Step 2: Add knowledge in a dedicated folder

  1. Go to Knowledge -> Sources.
  2. Create/select folder: dogs/.
  3. Upload dog-care docs (feeding, age stages, vaccination schedules, emergency symptoms).
  4. Start sync and wait for completion.

Open Dashboard → Knowledge → Sources to manage folders and uploads (same navigation as Widget under Dashboard → Widget).

Step 3: Configure tools and channels

  1. Enable core tools (search/retrieval, optional web lookup).
  2. Connect one channel for testing (web widget or inbox channel).
  3. Keep advanced integrations disabled for first validation pass.

Step 4: Test your first conversations

Run these prompts:

  • "My 3-month-old puppy has loose stool, what should I do right now?"
  • "Create a weekly feeding and walk plan for a 2-year-old labrador."
  • "What vaccines are typically needed in year one?"

Expected behavior:

  • Responses should cite practical steps and safety disclaimers.
  • Unknowns should trigger clarification questions.
  • Sensitive/medical edge cases should route to professional advice.

Step 5: Upgrade from demo to production

  • Add folder-level permissions for team access.
  • Add escalation workflow for high-risk questions.
  • Add integrations (calendar/CRM/ticketing) once baseline quality is stable.

Troubleshooting

  • If answers are weak: re-check folder sync status and source quality.
  • If tool calls fail: verify tool auth keys and retry with one tool enabled at a time.
  • If latency is high: reduce tool fanout and use model routing defaults.

GitHub & engineering assistants

Ground agents on internal docs and call GitHub MCP tools for commits, files, and repo context—without pasting tokens into chat.

Tool calling & agentic execution

How WisebotAI runs tools—built-ins, custom tools, MCP integrations, sandboxes, and the execution pipeline.