● Platform · Artificial Intelligence

Design how your agent thinks.
No code.

Prompt Manager (AI Configuration) is the segmented editor where you define your agent's role, style, rules and knowledge. Templates, variables, documents and AI-assisted improvement — all in one place, with live preview and automatic versioning.

5,000context characters per agent
+12ready-to-use templates
0lines of code required
MI-IA
+
Connected👁 Preview🤖 Chatbots
Prompt Configurator
Create effective prompts for better AI responses.
≡ Segmented📄 Single prompt
4898 / 5000
Prompt Segments
✨ Improve+ Segment
ticketing↑ ↓ 🗑
Role↑ ↓ 🗑
General style↑ ↓ 🗑
WhatsApp instructions↑ ↓ 🗑
Web instructions↑ ↓ 🗑
Why it matters

A poorly designed prompt
is an agent that doesn't work.

The problems we see in every team operating conversational AI. Prompt Manager solves them by design.

01

Monolithic prompts

3,000-word blocks with no structure. Impossible to maintain. Changing tone breaks business rules.

02

No versioning

Someone edits, something breaks, nobody knows what changed. No rollback. Production and staging out of sync.

03

Outdated context

Old docs in the RAG. Hardcoded variables. Schedules and prices in text nobody updated.

04

Invisible errors

Redundancy, ambiguity, contradictions. The model gets confused and nobody notices until a customer complains.

05

Every agent from scratch

No templates. Every new bot is a project. The team reinvents the wheel every time.

06

No collaboration

Business writes in a Google Doc, tech copies and pastes. Edits not applied, contradictions between revisions.

07

No way to test

You change the prompt and pray. No live preview, no testing before affecting real production.

08

Missing multi-channel

WhatsApp, web and IG need different tones. Without conditional segments, you end up with 3 bots and 3 duplicate prompts.

How it works

Segmented editor. Live preview. AI that helps you iterate.

Segmented editor

Your prompt in blocks you can move, edit and reuse.

Instead of an endless text block, organize agent behavior in segments: Role, Style, Channel instructions, Rules, Exceptions. Each is edited separately, ordered with drag & drop, and automatically compiled to the final prompt the model receives.

  • Independent and reorderable segments
  • "Segmented" or "single prompt" mode as you prefer
  • Live token and character counter
  • Automatic compilation to the model
  • Preview with the final rendered prompt
  • Exportable in XML format for integrations
MI-IA
+
Connected👁 Preview🤖 Chatbots
Editor
ticketing
Check ticket: When the user wants to know the status of a ticket, use find_ticket with the tracking_code. If they don't have it, ask for it.

Create ticket: When the user wants to open a ticket, use create_ticket. Collect name and email/phone before calling the tool.
Role↑ ↓ 🗑
General style↑ ↓ 🗑
WhatsApp instructions↑ ↓ 🗑
👁 Preview
# ticketing
Check ticket: When the user wants ticket status, use find_ticket with tracking_code.
# Role
You are MIA, Databot's AI assistant. Your goal is to handle and resolve queries...
# General style
Friendly, professional tone. Concise responses, no filler phrases...
compiled live● synced
Variables and documents with @

Native RAG. Connect documents, variables and actions with @.

Type @ and the menu appears: your loaded documents (PDFs, FAQs, catalogs), context variables (date, timezone, user name, last order) and available actions (check ticket, create lead, schedule). The agent accesses in real time, not a static dump.

  • PDF, DOCX, URL, sitemap documents
  • Variables: user, channel, session, custom
  • Tools Manager actions listed inline
  • Sync in seconds when uploading a doc
  • Semantic search over your KB
MI-IA
+
Connected👁 Preview
Create ticket: When the user wants to open a ticket, use create_ticket. Collect name and email before calling.
If the user provides full name, separate into name and last_name. @
📄
@Knowledge base MIA · Databot.pdf
Document · 11.11 MB
@Product Recommendations
Document · Custom
@custom_datetime
Current date and time
@timezone
User timezone
💡 Type @ to insert documents, variables or Tools Manager actions
Pre-built templates

Ready templates. Sales assistant, multichannel and more.

Each template comes with pre-assembled segments and suggested variables. Sales Assistant brings 5 segments for support + recommendation + referral. Multichannel Assistant brings 6 segments with conditionals that adapt behavior by channel (formal WhatsApp, casual Web, conversational IG).

  • Templates by vertical (sales, support, scheduling)
  • Conditional segments by channel
  • Suggested variables ready to replace
  • Customizable before applying
Prompt Templates
Select a template to get started quickly
×
🛒
Sales Assistant
Ideal for customer service, product queries and referrals to agents.
≡ 5 segments@ With variables
💬
Multichannel Assistant
6 segments with conditionals by channel: WhatsApp, Web and Instagram.
≡ 6 segments@ With variables
🎯
Technical Support
Escalation, SLA and automatic ticketing referral segments.
≡ 4 segments@ With variables
Measurable impact

What changes when you operate AI with Prompt Manager.

-80%
Setup time
for a new agent from template vs. from scratch.
3x
Iteration speed
with live preview and AI-assisted improvement.
100%
Versioning
of every change, with author, date and 1-click rollback.
+24%
Response quality
after passing redundancy/ambiguity analysis.
1
Workspace
to operate multiple agents (MI-IA, Sales, Support, etc.).
0
Lines of code
required to design and publish a prompt.

Ready to operate your AI with structure?

We'll show you in 30 minutes how to configure your first agent with Prompt Manager — from a template, with your documents and variables.

Book a meeting