Legacy systems and IT queues: what's holding back AI in companies?

Carla is the IT manager of a medium-sized distribution company. The company's ERP system is over a decade old, runs on a local server, and has its own unique way of behaving: on Mondays, backup routines slow everything down; on Fridays, someone always needs to restart something.
IA 5 min read By: Skyone

Ticket number 214: How legacy systems and IT queues are hindering AI

Carla is the IT manager of a medium-sized distribution company. The company's ERP system is over a decade old, runs on a local server, and has its own unique way of behaving: on Mondays, backup routines slow everything down; on Fridays, someone always needs to restart something.

In Carla's queue, ticket number 214 has been waiting for two weeks. It belongs to Paulo, the sales director, who requested a report cross-referencing sales and inventory. Carla's team of five was busy dealing with a server outage.

Paulo wasn't one to wait around. He exported the data to a spreadsheet, uploaded it to an AI tool, and put together the report in an afternoon. Carla only found out three weeks later, by accident.

Neither of them was wrong. The problem was what was underneath.

(Carla and Paulo are fictional characters, inspired by the behavior described in the research.)

Legacy systems and an overburdened IT infrastructure are the biggest obstacles to AI in companies. In the Vibe Coding Brazil Panorama, only 22% of companies use AI or LLMs in their development cycles, and legacy ERP systems and a lack of APIs hinder scalability. When IT takes anywhere from 2 days to more than 2 weeks to respond, 73% of users have already created their own solution.

Accelerated adoption, aging infrastructure

The Panorama report uses a summary that encapsulates the scenario: accelerated adoption, aging infrastructure. Over 80% of companies plan to increase their AI and automation budget in the coming year, but only 22% already use AI or LLMs in their development cycles.

What keeps the scale down? Three things, in the words of the report:

ProblemWhat it means
Legacy ERPCentral systems and legacy data
No APIsThere is a lack of bridges connecting us
IA strandedApplications that fail to scale

An AI that fails to handle revenue, inventory, and HR is far less useful than promised.

How much does the wait cost?

This was precisely the case with ticket 214. The study cross-referenced IT response time with the creation of in-house solutions:

Scenario (environment with incentives for AI)They created a solution on their own
IT responds in 2 days to more than 2 weeks.73,0%
IT responds in less than 1 day.56,4%

And 85.3% of creations happen individually or by the direct team, without IT. The long queue creates shortcuts, and IT, busy putting out fires, has no time on the agenda for governance.

Is your legacy holding back AI? Six questions

  1. Is the most important data stored in a system without APIs?
  2. Does extracting a new report require exporting spreadsheets?
  3. Does IT spend more time fixing problems than improving processes?
  4. Do backups and resource-intensive routines affect day-to-day performance?
  5. Has IT ever discovered parallel automation only after it already existed?
  6. Has any AI project been delayed due to "lack of access to data"?

Three or more "yes" answers suggest that the bottleneck is at the base.

What changed in Carla's routine?

Months later, the distributor migrated its ERP system to the cloud. That's how the story changed.

First, the foundation. Skyone Autosky moved the ERP system to a managed cloud without rewriting the software code. The report describes the chain as follows: migration (complex ERPs and legacy systems move to the cloud), stability (guaranteed high operational availability), IT liberation (moving out of "firefighting" mode), and innovation (new projects within institutional rules).

Then, the data. With Skyone Studio, ERP data is no longer bogged down. iPaaS connects systems and eliminates silos, Lakehouse unifies trusted data, and orchestration puts automations, data, and AI agents in the same flow, with governance from the source.

When Paulo asked for the next report, the conversation changed. It wasn't "when can IT do this?", but rather "what's the best way to do this securely?". Paulo's report was created within the rules, with no spreadsheets straying off the beaten path.

(Carla's outcome is illustrative, not a measured result in a client.)

From pain to solution: how Skyone helps

ChallengePanorama DataSkyone Solution
Legacy hinders AI scalability22% use AI in developmentSkyone Autosky: Managed cloud ERP, free IT to innovate.
Silos and fragmented integrations60.9% without data guidelines (fragmented group)Skyone Studio: iPaaS, lakehouse and orchestration
Self-creation73.0% when IT takes longerAutosky + Studio: stable base and accessible data, within the rules.

Going back to ticket 214

When someone in your company creates a shortcut, it's rarely rebellion. It's the response of someone who needs to work and has encountered an obstacle. If you want AI to truly advance, look first at what's underneath: accessible data and an IT department with time to think.

Download Panorama to see the complete scenario.

Skyone
Written by Skyone

Start Your Digital Transformation Today

Transform Your Business with Skyone. Request a demo or schedule a call with our experts to discover how Skyone can accelerate your digital strategy.

Subscribe to our newsletter

Stay up to date with Skyone content

Contact Sales

Have a question? Talk to a specialist and get all your questions about the platform answered.