Inbox Central · AI Agents · Kortex

Your support operation, on one platform.

BotDynamics brings your conversation channels together in Inbox Central. Your AI agents reply with your business knowledge, and your team steps in when it matters.

0
Agents on duty
0
Channels
agent://lucy
Example
Books, charges and escalates when needed Sales agent
WhatsApp Business API Facebook Messenger Instagram Direct Telegram Webchat Voice SMS Email
[ The platform ]

Seven pieces, one operation.

These are not separate products: each piece rests on the one before it. Channels feed the inbox, agents work on those conversations, and what happens there is what you can build on and measure.

01

Inbox Central

WhatsApp, Messenger, Instagram, Telegram, webchat, voice, SMS and email land in one inbox, where your team works alongside your automations.

02

AI Agents

A router directs, a knowledge agent answers and a flow agent runs the process. They hand off to a person when it matters.

03

Knowledge

Your configured sources feed the agents' answers. When the evidence falls short, the agent says so or hands off.

04

Kortex Builder

You describe the process and Builder proposes the flow, validated against the resources you already have. Applying, saving and publishing are your actions.

05

TeamChat

Internal chat with groups, images and documents, so your team coordinates without leaving the platform.

06

Kortex

Ask your operation in plain language: indicators, trends and conversations that need attention.

07

Marketing

Messaging campaigns over your contact lists, with the status of every send on record.

Builder and Kortex work on what already lives in the platform: Builder builds with configured resources and does not create integrations on its own, and Kortex queries the operation hosted in BotDynamics. That is why the pieces come in this order.

[ 01 — Inbox Central® ]

All your channels in one inbox.

Stop jumping between tabs. Messaging, voice and social media converge in a single interface for your whole team.

Inbox Central
Example
WA Laura M. WhatsApp
Handled by: Ana
Hi, I placed an order and I'd like to change the billing details.
Of course. To change the billing on an order that's already issued, I'll pass you to someone on the team.
Lucy · AI agent · 10:41
Handed to Ana · full history
Hi Laura, I'm Ana. I can see your order: send me the new tax ID and company name and I'll fix it now.
Ana López · 10:42
One inbox · several channels · AI and team Inbox Central

Example with synthetic data. It does not reflect any customer's operation.

Messaging

Every conversation lands in the same inbox, whichever channel it comes from.

Voice

Receive calls on the same platform, with the conversation's context.

Multichannel

WhatsApp, Messenger, Instagram, Telegram, webchat, voice, SMS and email.

One single inbox

Your team handles everything from the same place, with each conversation's history.

Automation

Your agents resolve the repetitive work and hand your team what needs a person.

Analytics

Kortex answers about this operation: indicators, trends and reports.

[ 02 — AI Agents ]

Agents that answer, route and run processes.

A router identifies the need and directs the conversation. A knowledge agent answers from your configured sources. A flow agent guides questions, forms and handoffs according to the logic you define.

01

Reception, support and sales

One agent greets and directs, another resolves support questions and another guides the sale up to the payment link. You define what each one does and with which resources.

02

WhatsApp Calling

Voice calls inside WhatsApp, in the customer's same conversation.

03

Guided processes

Books appointments, collects data with forms and sends the payment link, following the flow you define and the resources you have configured.

04

Escalates with context

When the customer asks or the evidence falls short, the conversation moves to your team with the full history.

05

Control over what gets published

Every agent version records who published it and when, and nothing ships without a human review.

06

Contact reasons

Conversations are grouped by what prompted them: what your customer asks for and how often.

[ 03 — Knowledge ]

Answers grounded in your business knowledge.

You organize your information sources so your agents find relevant content and ask for help when the evidence falls short.

01

Your sources, versioned

You upload files and web addresses. Every update creates a knowledge version, so you know what the agent answered with and when it changed.

02

Hybrid retrieval

Retrieval combines semantic and text search, then ranks the results before answering. This is not keyword matching.

03

Answers with citations

Every answer is validated against the retrieved sources. If the citation does not hold, the answer is not sent.

04

Answer, clarify or hand off

These are the only three possible outcomes. When the evidence falls short, the agent asks or passes the conversation to a person instead of improvising.

05

Side questions without losing the process

If the customer asks something outside the process under way, the agent can answer and resume where it left off. Enabled by configuration; not on by default.

06

Controlled publishing

A version is not published if its configuration was not evaluated or if sources remain unprocessed under a different profile.

Uploading documents is not training or fine-tuning a model: the agent consults your sources at answer time. The scope of the answers depends on the sources you have configured.

Knowledge base
Example
Configured sources
  • returns-policy.pdf v3 · active
  • shipping-guide.pdf v2 · active
  • store.example/faq v5 · active
Can I return an online purchase after 30 days?
Yes. For online purchases the window is 45 days, as long as the product is unused and in its original packaging. [1]
[1] returns-policy.pdf · v3 · p. 4
Answer ✓ Clarify Hand off
Answer validated against the source Knowledge agent

Example with synthetic data. It does not reflect any customer's operation.

[ 04 — Kortex Builder ]

Describe the process. Build the flow.

You turn your instructions into flow proposals, review their technical validation and test every path before publishing it.

Kortex Builder agent: orders_support
Example
Ask for the order number, check its status with the checkOrderStatus operation and, on error, hand off to the general_support queue.
Proposal +3 nodes · +4 connections · 1 warning
+ Question · order_number
+ Operation · checkOrderStatus
+ Handoff · general_support queue
! No timeout handling — to confirm

Verified resources: checkOrderStatus operation ✓ · general_support queue ✓

Apply to design Simulate Publishing requires a human action
Unsaved · unpublished Technical validation

Example with synthetic data. The proposal is not applied or published on its own.

01

You describe, you don't draw

You explain the process in plain language inside the flow designer and Builder proposes the matching flow.

02

Proposals with a diff

Every proposal arrives with what changes, the resources it uses and the warnings it found. You review it before anything is applied.

03

Only with what you already have

It builds with your configured API operations, queues, forms and payments. If a resource is missing it says so and stops: it will not invent a queue that does not exist.

04

It asks when a decision is missing

Faced with an ambiguous instruction it asks for clarification instead of inventing dates, times or business rules.

05

Simulate before publishing

You test the happy, invalid and error paths in the simulator, with API and payment doubles.

06

Publishing is yours

Applying, saving and publishing are human actions. Builder never publishes on its own, and every version records who published it and when.

Builder does not manage the runtime agent's identity or model, does not create integrations or resources, and does not query your customers' data. Usage is subject to per-account configurable limits.

[ 05 — TeamChat ]

Resolve with your team, inside the platform.

Internal chat for agents and supervisors to coordinate while they work, without leaving the platform.

Work groups

With members and roles defined by your account.

Direct messages

Between agents and supervisors, one to one or in a group.

Shared files

Images and documents inside the chat.

Notifications

Browser alerts, with adjustable sound and volume.

Sales Team
Example · 12 members
JM
Can someone help with the customer on account #1247?
Juan Martínez · 10:23
AL
I'll take it, looking at it now.
Ana López · 10:24
MR
Perfect. I'll share the updated document.
María Rodríguez · 10:25
Carlos is typing
[ 06 — Kortex ]

Ask your operation in natural language.

Kortex lives inside your admin panel. Ask in plain English and it answers with your operation's data: indicators, channels, team and conversations that need attention.

01

Natural language queries

"How many conversations did we close yesterday?" or "Which channel had the most?" — with a breakdown by channel and status.

02

Trends over time

Compare how your indicators evolve between periods and surface what needs attention.

03

Team performance and reports

Compare your agents' performance, check your bots' health and generate the period's report from the same conversation.

04

Every channel, one reading

Kortex queries the whole operation hosted in BotDynamics: WhatsApp, social and voice in one reading.

Kortex Inbox Central
Example
How was yesterday?
Daily summary Sample data
conversations0
saved0
completed0
waiting0

Top of the day
WhatsApp › Sales — 560 (47%)

Example with synthetic data. It does not reflect any customer's operation.

Campaign · Black Friday
List of 240 contacts
Example
0
Contacts
0
Delivered
0
Read
Preview

Exclusive Black Friday!

Up to 50% off the entire catalog. Today only.

Segment Example filter
[ 07 — Marketing ]

Messaging campaigns, with a record of every run.

You define the message, choose the channel and the recipient list, and follow the campaign's status from start to finish.

01

Recipient list

You build the campaign on a contact list, with filters over your base.

02

Configurable message

You define the message the campaign goes out with.

03

Send pacing

You set the spacing between messages so the channel is not flooded.

04

Status of every send

Sent, delivered, read and failed, campaign by campaign, with total duration.

05

Channel rules

Sends go out through the configured channel, subject to its template and consent policies.

[ Questions ]

What people usually ask us.

Four answers before the demo, so you know exactly what you are looking at.

01

What is the difference between Kortex and Kortex Builder?

Kortex answers questions about your operation: indicators, channels, conversations that need attention, team performance and reports. Kortex Builder does not query data: it lives inside the flow designer and turns your instructions into flow proposals that you review and publish. One reads the operation, the other builds the process.

02

How does a person take part?

At three points. The agent hands the conversation to someone on your team when the evidence falls short or when the customer asks, and that person picks it up in Inbox Central. In Kortex Builder, applying, saving and publishing a flow are always human actions. And in TeamChat your team coordinates while it works.

03

What does an integration need?

That the resource already exists and is configured: the API operation, the queue, the form or the payment method. Kortex Builder builds on those resources, but it does not create or connect them on its own. If one is missing, it says so and stops rather than inventing it.

04

Which channels are available?

WhatsApp, Facebook Messenger, Instagram, Telegram, webchat, voice, SMS and email. Each channel is configured separately and they do not all offer the same features, so in the demo we review which ones apply to your operation.

[ Start today ]

Let's see your operation running.

Book a 30-minute demo and we'll show you how your channels, your agents and your business knowledge are configured.

Message us