AI Services · ACL powered by DataArt

Artificial intelligence for enterprises, integrated into your operations

We build and operate AI assistants, agents and models that work with your systems and data, with human review on critical decisions and traceability for every response.

ISO/IEC 27001:2022
ISO 9001:2015
In short

What is ACL's artificial intelligence for enterprises service?

The artificial intelligence for enterprises service from ACL (ACL powered by DataArt), a Chilean technology services company in business since 1991, builds and operates AI assistants, agents and models connected to the client's systems.

Every project starts from a prioritized business case, is delivered as a tightly scoped MVP and includes human review on critical decisions.

How do I implement AI in my company with a vendor?

Start with a focused, measurable business case, not with the tool.

ACL prioritizes it in an assessment, builds an MVP connected to your systems and takes it to production with human review, access control and traceability. If the problem can be solved with RPA or an integration, we'll tell you.

What we do

  • Assistants that answer in natural language based on your documents and data.
  • AI engines that evaluate and standardize decisions, reviewed by an analyst.
  • Integration of artificial intelligence with the systems and applications your company already uses.
What we build

AI solutions we build

We choose the type of solution based on the business problem to be solved. ACL has delivered projects of all three types.

Assistants for your documents

They answer in natural language based on your repositories and integrate with the platform your team already uses.

An assistant for an analytics platform was delivered in one month.

AI that evaluates and standardizes decisions

It applies the same criteria to evaluations that currently depend on each analyst. A person reviews and validates the result, and every decision is logged.

AI inside your applications and workflows

We add AI models to existing apps and processes. We combine low-code (Power Automate) with custom development (React, Flutter), depending on what each case calls for.

Looking for AI for software testing? See generative AI applied to testing.

Process

How we implement AI in your company

Five steps, from business case to operation.

  1. Assessment and prioritization

    We identify where AI moves a business metric and whether your data is ready. If you need an AI-ready data platform, we address it here. If the case calls for automating rule-based tasks, we handle it with our process automation with RPA and AI.

  2. Focused MVP

    We build the first version around a specific case, connected to your real systems from day one.

  3. Validation with the business team

    The business team tests the solution with real cases and defines which decisions require a person to review the result before it is considered valid.

  4. Go-live

    We deploy with per-user access control and logging of every response, so results can be reviewed later.

  5. Operation and metrics

    We measure usage and results against the metric we defined in the assessment, and we fine-tune the solution based on that data.

AI agents

AI agents for enterprises: what they need to operate

An AI agent takes actions in your systems, like opening a ticket or preparing a document. That's why it needs more control than an assistant that only answers.

What an agent requires

  1. Least privilegeIt only accesses what its task requires.

  2. Human approvalAt critical steps, before the action takes effect.

  3. A log of every actionWhat it did and with what data.

At ACL, we apply that governance to the way we build software with AI ourselves: SPIREX®, our agentic orchestration framework, puts AI agents at every stage of the development lifecycle, with human approval (Human-in-the-Middle) at every transition and a log of every run.

Artificial intelligence in enterprises: case studies

ACL projects with real timelines and teams.

See more ACL case studies
Analytics · international operation

Natural-language assistant for a document repository

Result

Queries answered without manual searching. Delivered in 1 month, in 2 milestones.

Analytics · international operation

Natural-language assistant for a document repository

Result

Queries answered without manual searching. Delivered in 1 month, in 2 milestones.

Challenge
Users searched a large document repository by hand. Finding an answer was slow.
Solution
A virtual assistant with natural language processing that searches the repository and drafts the answer, with a new interface integrated into the existing platform. AI services on Azure and OpenAI.
Higher education

AI engine that evaluates and standardizes decisions, with human review

Result

Standardized criteria, less time spent on analysis and traceability for every decision. 3 months, team of 4 professionals.

Higher education

AI engine that evaluates and standardizes decisions, with human review

Result

Standardized criteria, less time spent on analysis and traceability for every decision. 3 months, team of 4 professionals.

Challenge
Evaluating resumes for faculty positions was manual and inconsistent: each evaluator applied different criteria.
Solution
An AI engine that evaluates education, experience and skills by discipline and by course, with review and validation by analysts, reporting and traceability. Power Automate, React and Azure OpenAI.
Higher education · sustainability

AI inside a mobile app (image recognition)

Result

A tool that optimizes water use and promotes native species. 3 months, 4 professionals.

Higher education · sustainability

AI inside a mobile app (image recognition)

Result

A tool that optimizes water use and promotes native species. 3 months, 4 professionals.

Challenge
There were no digital tools for designing sustainable landscaping adapted to the climate and soil of each site.
Solution
A mobile app with recommendations based on climate and soil, water consumption calculations, maps and AI-powered plant recognition. Flutter, Firebase and Cloud Vision API.

Have a business case in mind? We'll prioritize it with you.

Schedule an AI assessment
Security

Security and control of your data

The first thing an IT committee asks is what happens to the data. Here's how we handle it in AI projects.

ISO/IEC 27001:2022 · RI 27000-88ISO 9001:2015 · RI 9000-9942Certified by IRAM, with IQNet recognition
  • Human review

    In decisions that affect people or the business, an analyst reviews and validates the AI's output before it is used.

  • Traceability

    In those processes, every response and every decision is logged, with reporting so it can be reviewed later.

  • Access control

    Access to the solution and to the data it queries is controlled per user.

  • Certified security

    ACL operates an information security management system certified to ISO/IEC 27001:2022 and a quality management system certified to ISO 9001:2015. See our ISO 9001 and ISO/IEC 27001 certifications.

Backing

ACL, in Chile since 1991, is part of the DataArt group

  • 6,000+engineers and consultants · DataArt group
  • 40+locations · DataArt group
  • 80NPS · DataArt group, 2025
  • 95%client retention · DataArt group, 2025

Companies that trust ACL

  • Esmax
  • Nexus
  • Rheem
  • Sparta
  • Deloitte
  • Blue Express
  • Oxiquim
  • AFC Chile
  • BTG Pactual
  • Banmédica
  • Taivo (formerly Inktavo)
  • Antofagasta Minerals
  • Caja Los Héroes
  • Accenture

Frequently asked questions about artificial intelligence for enterprises

How does ACL run an AI implementation project?

In five steps: assessment and prioritization, a focused MVP, validation with the business team, go-live, and operation with usage metrics. The assessment pinpoints where AI moves a business metric, and the MVP connects to your real systems from day one.

How long does an artificial intelligence project take?

The three AI projects shown on this page took between one and three months. An assistant that answers questions about a document repository was delivered in one month and two milestones, and an AI engine for evaluating applicants in three months with four professionals.

The timeline depends mainly on how many systems need to be integrated and on data quality. The assessment defines the scope and timeline of the MVP before we build.

What is an AI agent and how is it different from a chatbot?

An AI agent takes actions on your systems, like creating a ticket or preparing a document, within the permissions you give it; a chatbot converses and provides information, but doesn't act.

That's why an agent needs more governance: least privilege, human approval at critical steps and a log of every action. In our own software development, that governance comes from SPIREX®.

How do you keep the AI from making up answers?

By having the assistant answer based on your documents and data, and by adding human review to decisions that affect people or the business. In those processes, an analyst validates the result before it is accepted, and every response is logged so it can be audited.

How do you protect my company's data in an AI project?

With per-user access control, logging of queries and an information security management system certified to ISO/IEC 27001:2022, registration RI 27000-88. That certification is audited by an external body.

For reference, the document assistant and the evaluation engine shown on this page run on Azure AI services.

Who do I talk to at ACL to evaluate an AI project?

ACL's AI team: schedule an AI assessment. In an initial technical meeting, at no cost, we review the business case, the systems to be integrated and the available data.

From there, we define the scope and timeline of the MVP, with the human review each decision needs.

Talk to ACL's AI team

In an initial technical meeting, at no cost, we'll review:

Your caseThe available dataWhether AI, automation or both is the right fit
Schedule an AI assessment