ZRN Dev

zrn.dev

AI Agents · RAG · Automation · Data & ML

// est. 2017

// operating — zrn.dev

Hola.

Your operation, on autopilot.

We build AI agents, automations, and data platforms that work 24/7 for your company — measurable, supervised, and in production, not in demos.

// What are you looking for?

Automate processes Use my data Build my platform
Book a call See results

Reply in under 24h — no strings attached. · info@zrn.dev

In action

What a ZRN agent working looks like.

zrn — agent: run 
$ zrn agent run --task "reconcile invoices"
✓ plan ................. 3 steps defined
✓ rag .................. 248 docs · 12 relevant
✓ tools ................ ERP + bank + email
✓ validate ............. schema OK · conf 0.94
● done ................. 38 invoices · 0 errors
  cost $0.42 · 41 s · human reviewed 2 cases

// in plain words: it reconciled 38 invoices in 41 seconds · cost $0.42 · escalated 2 edge cases to a person

−73%

incident resolution time

40%

of tickets resolved without humans

4 wks

from pilot to production

8+

years building systems

Companies that trust us

A sample of the organizations we’ve accompanied with intelligent platforms, AI, and data.

Services

What we can be running for your company in a matter of weeks.

ai agents

Your operation on autopilot

Agents that execute the complete job: reconcile, answer, classify, and operate your systems while your team decides.

rag

Answers from your own data

AI that reads your documents and databases to answer with precision — source cited, nothing made up.

automation

Processes that run themselves

Integrations, alerts, and reporting that stop eating your team’s hours. From repetitive task to complete pipeline.

data & bi

Decisions made on data

Dashboards and operational reporting on your data: what’s happening in the company, visible and current, without waiting for month-end close.

machine learning

See what’s coming

Demand, risk, and behavior forecasting with models trained on your data and measured against your reality.

web platforms

Your platform, built to fit

Full-stack web development with our usual engineering rigor — and AI inside from the design.

How we work

From diagnosis to production in weeks, not months.

  1. 01 · 1 week

    Diagnosis

    We find the processes where AI pays for itself, with numbers.

    workshops · process analysis · business case

  2. 02 · 4 weeks

    Pilot

    An agent working on your real data, with a measurable result.

    tight scope · metrics agreed upfront

  3. 03 · ongoing

    Production

    We operate, measure, and scale what works.

    supervision · cost per task · continuous improvement

Portfolio

Real cases. Results that matter.

ZRN 360º

Our product2026
https://zrn360.io

360º platform for running companies end-to-end: CRM, ERP, HR and accounting in one system.

+edit
users
99%
uptime
<edit ms
p95
AstroTypeScriptTailwind
Own productSaaSPlatformData

OpsPilot

2025

AI ops copilot that cuts incident triage time by 73% by integrating logs, tickets and runbooks.

−73%
triage time
24/7
alerts
+40%
auto-resolved
OpenAILangChainTemporalGCP
AI AgentsLLMGCP
Dashboard predictivo IFES

Dashboard predictivo IFES

2025

Predictive dashboard for IFES: dropout forecast, academic risk and operational load with custom models over LMS data.

+edit
users
<edit%
accuracy lift
24/7
monitoring
Pythonscikit-learnPostgreSQLMetabase
MLPredictiveEducation
Moodle IFES

Moodle IFES

2024

Moodle setup, configuration and ongoing maintenance for IFES. Students, courses, assessments and reporting without friction.

<edit
incidents/month
<edit h
avg response
99.9%
uptime
MoodlePostgreSQLPHPRedis
MoodleLMSSupport

FinPay Core

2025

Payments platform processing 40k tx/day with p99 latency < 120 ms and full audit trail.

40k+
tx/day
p99 <120 ms
latency
100%
auditable
PostgreSQLRedisOpenTelemetryAWS
TypeScriptPostgreSQLAWS

RetailSync

2024

Real-time inventory sync across 300 stores; duplicate orders driven to zero.

300
stores
0
duplicates
<1 min
sync
GoKafkaClickHouseK8s
GoKafkaK8s

Tech stack

Technology chosen on merit to build AI that won't let you down.

FAQ

How much does a project cost? +

It depends on scope. We start with a fixed-price discovery phase and then work in weekly milestones.

What if the AI agent makes a mistake? +

Every agent validates its output against a schema, reports its confidence level, and escalates to a human below a threshold. It acts within defined limits, always.

Is our data safe? +

We run RAG on your own infrastructure when needed: your documents never leave your environment, and every access is audited.

Do you offer maintenance? +

Yes. 80% of our clients keep a maintenance and continuous evolution retainer.