Build intelligent rag powered agents in

minutes the secure no code way with

kicki.

Whether you're building an agent for

healthcare finance education or

customer support, KI helps you easily

create, test, and launch smart AA agents

that give accurate and contextaware

responses all through a simple,

intuitive interface.

[Music]

Once you log into the Kiki platform,

you'll be directed to the Kicki

dashboard.

On the left panel, you'll find the

option called Kicki Flow. Click on it.

This will take you to the KIKI flow

builder.

Now, let's create a new chat agent. You

can either open an existing project or

create a new flow by clicking on new

flow. You'll see three flow types. basic

prompting vector store rag simple agent.

You can also start from a blank flow for

full customization.

For this demo, we'll select vector store

rag. Once selected, you'll see a default

rag flow appear. Rename the project

using the edit icon beside the title if

you'd like. You can keep the default

setup or enhance it by dragging new

components from the left panel or even

build your own Pythonbased custom

components for advanced logic.

The rag flow has two stages, indexing

and querying. In the indexing stage,

you'll load your data files into a

database. Here we're using Chroma DB for

vector storage and retrieval. I've also

created a custom component that

automatically loads data directly from

Google Cloud Storage. Enter your open

AAPI key which is used for generating

embeddings to store your data as vectors

in Chroma DB.

Click the play button on the database

component to store your data.

Whenever you update data, reselect your

repository name and rerun the DB

component to refresh.

After indexing, move to the querying

stage. Here the user's query is embedded

into vectors and the system retrieves

the most relevant results from Chromma

DB.

You can also customize the prompt to

suit your use case.

[Music]

Once everything is set, head over to the

playground to test your chat agent. You

can interact with it just like a real

user. Ask questions, verify retrieval

accuracy, and fine-tune your prompt for

better results.

Once done, click save flow to store your

design.

Now let's connect your chat agent flow.

On the right side of the board, click

your profile icon to open the settings

page. On the left, select kickflow API.

Then click generate API key.

[Music]

Copy and securely save your key.

[Music]

To give your chat agent its look and

feel, go to channel configuration. Here

you can view your existing chat agents

or create a new one.

Click chat agent to start branding your

agent. Customize its name, logo, color

palette, and personality to match your

brand identity. Next, paste the API key

you generated earlier to connect your

chat agent flow.

Click save and your chat agent is now

live and ready to interact with users.

To see the chat agent you've designed,

simply copy the URL provided and open it

in your browser.

Enter your credentials and you'll be

redirected to your chat agents

interface. Here you can explore the chat

agents look, feel, and responsiveness

just as you designed in coli flow.

Watch your intelligent agent come to

life.

For example, I've created a flow related

to healthcare. In colli, you can build

agents for health care, finance,

education, customer support, and more.

You name it, you create it all through

your logic, creativity, and imagination

without writing a single line of code.

[Music]

[Music]

Once your chat agent is live, the next

step is understanding how it performs

behind the scenes. That's where Kicki

Lens comes in. Your built-in monitoring

system to trace, measure, and optimize

your agents performance in real time.

Click Holliki Lens from the left panel.

You'll land on the projects page where

you can create new projects or analyze

existing ones. This is your starting

point for all chat agent analytics.

Every flow, session, and metric begins

here. Tracing each coli flow run is

automatically recorded as a trace

capturing every step from user query to

final output. Each component prompts

retrievers or LLM calls appears as a

span in the timeline.

Tracing helps you debug, find slow

responses and optimize performance with

full visibility.

Sessions combine multiple traces from a

single conversation.

They let you review full chat histories,

analyze token usage, latency, cost, and

success rates all in one view. Perfect

for understanding user behavior and

improving consistency.

The dashboard is your visual control

center. Here you can monitor token

usage, latency and model cost. Track

success and failure rates. Compare

models and agents. Gain realtime

insights for optimization.

[Music]

It turns your data into clear actionable

visuals.

The home view shows all traces and model

usage in one place.

Filter by model or track overall

performance to monitor system health and

efficiency.

That's how you create a rag flow and

collick. Simple, powerful, and

completely no code. Keep building smart

agents with colli.

[Music]