botcampus R
Main Menu
botcampus R
Main Menu
Profile

Krivi AI Explained: How to Build Agents Without Coding

Krivi AI aims to solve a simple but painful reality:

Most people have ideas for AI agents, but
almost nobody wants to write Python, fight with APIs, or deploy servers.

“What if I could just describe the agent I want, connect a few tools, and have it work?”

That’s exactly the promise of Krivi AI — a way to build and run useful AI agents without writing code, using a visual editor, natural language, and ready-made integrations.

In this guide, we’ll walk through:

  • What Krivi AI actually is (in plain language)
  • The core concepts: agents, workflows, tools, memory, triggers
  • How Krivi AI works behind the scenes (simple mental model)
  • A step-by-step example: build a lead-qualification agent without coding
  • Real-world use cases you can ship today
  • Best practices so non-coders don’t accidentally create chaos

1. What Is Krivi AI?

At a high level:

Krivi AI is a no-code platform for designing, launching, and managing AI agents using a visual canvas and natural language — no programming required.

Instead of:

  • Spinning up servers
  • Wiring APIs manually
  • Writing complex code to call LLMs

You:

  • Describe what you want the agent to do in plain English
  • Drag & drop blocks (triggers, tools, logic) on a canvas
  • Connect your existing apps (Gmail, Sheets, CRM, Slack, etc.)
  • Hit “Run” and watch the agent work

Think of Krivi AI as:

  • Not just a chatbot builder
  • Not just “another automation tool”
  • But a digital agent studio where non-developers can build workers that:
    • Read data
    • Make decisions
    • Take action across tools
    • Run on schedules or events
    • Improve over time with tweaks and feedback

2. Core Concepts in Krivi AI (Beginner-Friendly)

You don’t need to know everything to start.
If you understand these five ideas, Krivi AI will feel much simpler.

Agents

An agent is your digital teammate.

You define:

  • Its role: “You are a Lead Qualification Agent…”
  • Its goals: “Score leads and decide who should be contacted today.”
  • Its tools: “You can read from this sheet, send Slack messages, draft emails…”

The agent uses AI + those tools to move toward the goal.

Workflows

A workflow is the path your agent follows.

Example:

  1. Trigger when a new row is added to a Google Sheet
  2. Read the lead’s data
  3. Ask the AI: “Is this a good-fit lead? Why?”
  4. If score is high → send a personalized email + notify sales in Slack
  5. If score is low → tag as “Nurture” and maybe add to a newsletter list
  6. In Krivi AI, you see this as connected blocks on a visual canvas.

Tools

Tools are how your agent touches the real world.

Some examples you might connect:

  • Gmail / Outlook → send emails
  • Google Sheets / Airtable / Notion → read & write records
  • CRMs (HubSpot, Pipedrive, etc.) → update deals, contacts, stages
  • Slack / Teams → send notifications
  • HTTP / Webhooks → talk to any API

You don’t write code to use these — you configure them via forms and dropdowns.

Memory

Memory lets your agent remember context over time:

  • What conversations happened before
  • What actions it already took
  • User preferences (tone, language, priorities)
  • Past decisions and results

Without memory, your agent acts like a goldfish.
With memory, it behaves more like a long-term assistant.

Triggers

Triggers define when your agent should wake up.

Common triggers in Krivi AI might include:

  • “New row added to Sheet”
  • “New email in this inbox”
  • “Webhook call from a form or app”
  • “Every day at 9 AM”

Once triggered, Krivi AI starts the workflow, passes data into the agent, and the magic begins.

Want to know about Ai Agents visit BotCampusAi-Workshop

3. How Krivi AI Works Under the Hood (Simple Mental Model)

You don’t need to see the source code — but a simple mental picture helps.

When your agent runs, Krivi AI basically does this:

  1. Perceive – Collects inputs: trigger data, records, emails, etc.
  2. Think – Sends a well-structured prompt + context to an LLM (AI model).
  3. Act – Uses the AI’s decision to call tools (send email, update CRM, etc.).
  4. Update State – Saves what happened into memory / logs.
  5. Loop or Finish – Either runs another step or ends the workflow.

You define what it should accomplish and which tools it’s allowed to use.
Krivi AI handles the boring parts: model calls, data passing, retries, error capture.

4. Example: Build a Lead Qualification Agent (No Code)

Let’s build a realistic agent you could actually use in your business:

“When a new lead is added to my sheet, qualify them and take the next step automatically.”

Step 1 – Clarify the Agent’s Job

Plain English description:

  • Reads new leads from Google Sheets
  • Scores them 1–5 based on fit
  • If score ≥ 4 → send warm, personalized email + Slack alert
  • If score 2–3 → tag as “Nurture”
  • If score 1 → ignore or log only

This description becomes the basis of your agent’s role prompt.

Step 2 – Create a New Agent in Krivi AI

Inside Krivi AI:

  • Click “New Agent”
  • Name it: Lead Qualifier – Google Sheet
  • Under Role / Instructions, paste something like:

“You are a Lead Qualification Agent.
For each new lead, read their data and score them from 1–5 based on how closely they match our ideal customer profile.
5 = perfect fit, 1 = bad fit.
Then choose what to do next:

  • 4–5: move to ‘Hot Lead’ branch
  • 2–3: move to ‘Nurture’ branch
  • 1: move to ‘Ignore’ branch.”

Krivi AI will use this as the agent’s “brain wiring”.

Step 3 – Add a Trigger: New Row in Google Sheets

On the visual canvas:

  • Add a Google Sheets Trigger block
  • Connect your Google account (no code, just login/permissions)
  • Select your Leads sheet (with columns like Name, Email, Company, Budget, etc.)

Whenever someone fills a form or you add a row manually, this trigger fires.

Step 4 – Send Data into the Agent

Next block: Krivi AI Agent (the Lead Qualifier you just created).

  • Connect the Sheets trigger output → Agent input
  • Map fields:
    • lead_name ← Name
    • lead_email ← Email
    • company ← Company
    • notes ← any notes field

The agent now receives structured data and can decide what to do.

You configure the agent to output something like:

 
{ "score": 4, "reason": "Matches ICP on company size and budget.", "recommended_path": "hot_lead" }

No coding — just selecting fields and defining output structure via UI.

Step 5 – Branch the Flow Based on Agent Output

Add a Conditional / IF block:

  • Condition: recommended_path == "hot_lead" → go down HOT LEAD branch
  • recommended_path == "nurture" → go down NURTURE branch
  • Else → IGNORE branch

Now your workflow visually splits into three paths.

Step 6 – HOT LEAD Branch: Send Email + Slack

In the Hot Lead path:

  1. Add an Email node (e.g., Gmail):

    • To: lead_email

    • Subject:

       
      Hey {{lead_name}}, quick idea for {{company}}

    • Body:

       
      Hi {{lead_name}}, I took a quick look at {{company}} and I think we can help you with {{short pain summary from agent}}. If you're open to it, I can share 2–3 specific ideas tailored to your situation. Best, Your Name

    (You can even let the agent generate short pain summary.)

  2. Add a Slack node:

    • Channel: #sales
    • Message:🔥 New hot lead scored {{score}}/5: Name: {{lead_name}} Company: {{company}} Reason: {{reason}}
    • Now sales hears about strong leads instantly.

Step 7 – NURTURE Branch: Tag and Save

  • In the Nurture path:
  • Add a Sheets Update node: write "Nurture" into a Status column
  • Optionally, add them to a newsletter list or send a soft intro email.

Step 8 – IGNORE Branch: Just Log

  • In the Ignore path:
  • Maybe just set Status = "Low Fit"
  • Or keep them for analytics only.

Step 9 – Test, Iterate, and Activate

  • Run the workflow with a few test leads
  • Inspect the agent’s scores and reasoning
  • Adjust the role prompt or criteria text if it’s being too strict/too generous
  • Once happy, switch the agent to active
  • You’ve just built a working AI lead qualification agent — without a single line of code.

5. Other Agents You Can Build in Krivi AI (Today)

  • Once you’ve built one, ideas start popping up everywhere. Some examples:
  • Inbox Triage Agent
    • Trigger: new email
    • Classify: urgent / important / newsletters / spam
    • Action: label, archive, draft replies for certain patterns
  • Customer Support Agent
    • Trigger: new support ticket
    • Read FAQs + past tickets
    • Propose a reply, auto-handle simple issues, escalate complex ones with a clean summary
  • Invoice & Expense Agent
    • Trigger: new invoice PDF in a Drive folder
    • Extract vendor, amount, category
    • Write to a sheet or accounting tool
    • Send weekly spend summary
  • Content Repurposing Agent
    • Trigger: new blog or YouTube link
    • Generate LinkedIn posts, X threads, email snippets
    • Save into a content calendar or schedule directly
  • Personal Productivity Agent
    • Trigger: every evening
    • Scan calendar + tasks + emails
    • Generate a next-day plan and email it to you

All follow the same pattern:

Trigger → Agent thinks → Agent chooses path → Tools act → State updates.

You’re just snapping blocks together on the Krivi AI canvas.

6. Best Practices for Non-Coders Using Krivi AI

A few guidelines so your agents are powerful and safe:

  • Start narrow
    Don’t automate your entire business on day one. Start with one small, annoying task.

  • Keep a human in the loop at first
    Let agents draft emails, updates, or actions — but review them before auto-sending.

  • Name everything clearly
    Nodes, branches, and agents with names like Classify Intent or Send Hot Lead Email are easier to maintain later.

  • Log and observe
    Periodically check what your agents are doing. Tweak prompts and rules based on real behavior.

  • Add guardrails
    Limit which tools an agent can use and what it’s allowed to change. For example, “never send refund emails automatically — only drafts.”

  • Version your agents
    When you improve an agent, treat it like a new version. If something breaks, you can roll back.

7. Conclusion: Krivi AI as Your No-Code Agent Studio

The old way of building automation:

  • Hire developers
  • Wait weeks or months
  • Pray everything stays in sync when tools change
  • The new way with platforms like Krivi AI:
  • Drag, drop, and describe what you want
  • Let AI handle the “how”
  • Iterate in days, not months
  • Give non-technical teammates real power over workflows
  • You don’t need to become a full-stack engineer.

You just need to understand:

  • What the agent should do
  • Which tools it’s allowed to touch
  • How to design clear, safe workflows
  • From there, Krivi AI can handle the heavy lifting.
  • For more breakdowns of agents, no-code workflows, n8n + LangChain + LangGraph stacks, and practical automations you can plug into your own projects, keep an eye on BotCampusAI — we’ll keep turning complex AI concepts into clear, buildable systems you can actually use.

Recent Post

Scroll to Top