Coding OS
← Portfoliocase study
Management with AI agents
We built the management web app for a software factory. Expected modules — clients, quotes, finance, projects, tasks — plus something else: AI agents working inside the flow.
Software factory
Platform · AI agents
3 agents in production
Context
The brief
They're not assistants that answer when asked. They run on their own, every day, and leave finished work in the system: loaded prospects, prioritized quotes, drafted proposals.
Challenge
The constraint
Move AI from a tool someone opens when they remember to part of daily commercial operations workflow.
deliverables
What we built
Module
Clients and prospects
Client and prospect management inside the platform.
Module
Quotes and proposals
Quotes and proposals connected to the commercial flow.
Module
Projects and tasks
Project and task tracking for the factory.
Module
Finanzas
Financial module integrated with operations.
IA
AI agents integrated into the flow
Three agents running on commercial operations on their own, every day.
outcome
AI stopped being a tool someone opens when they remember and became part of daily workflow.
process
How we did it
Agent 01 · Prospecting
Every day it scans Google Maps and search results for SMBs. It extracts data, visits each site, builds a digital situation and performance report, picks the top 3–4 prospects, loads them as leads and drafts a first outreach email.
Agent 02 · Commercial prioritization
It reviews all active quotes and prospects. It evaluates amount, margin and estimated effort, ranks the portfolio and indicates which client to follow up with first.
Agent 03 · Proposals
It triggers when a meeting ends and the transcript is generated. It detects what the client asked for and their pain point, proposes a solution and builds the proposal with brand design, saved in the app without prices or hours. After the meeting you only add numbers and send.
3
Agents running on commercial operations
The value
- An internal platform for a software factory
- Agents that run on their own and leave finished work in the system
- Prospects loaded, quotes prioritized and proposals drafted without waiting for someone to open AI
