There's a lot of talk about digital transformation, artificial intelligence, and bots
that solve everything. But when it comes time to implement them, many organizations
get frustrated.
Why? Because automating without understanding the process, without a rigorous
diagnosis or business judgment, is like building a highway on sand: it may look
modern, but it won't last or produce results.
Automate ≠ apply AI (let alone sell magic solutions)
At Novatium, we're clear: automation doesn't always involve artificial
intelligenceor complex platforms. Often, what's needed isn't an algorithm
, but rather a person with
process experience, who understands:
- how information flows
- where human time is wasted
- Scale up or down teams with agility
- and how to reorder tasks so that technology has a real impact.
The problem is that there are many promises that don't hold up:
Companies that claim to use AI when in reality they only connect forms to spreadsheets,
consulting firms that relabel simple automations as "machine learning," or tools that
solve a minimal part of the problem... but don't transform it.
Something that few consulting firms openly explain is that using artificial
intelligence has a cost, and not a minor one.
Each time an AI model—like ChatGPT, Claude, or Gemini—processes a query
it consumes tokens, a unit of measurement similar to words.
Platforms charge for every 1,000 tokens processed, both inbound and outbound, and those
costs can escalate very quickly if left unchecked:
- A chatbot with simple queries can cost $0.05 per interaction.
- An automated analysis of contracts or texts can cost tens of dollars per
document, depending on the model and volume.
- AI tools that integrate with RPA or data flows require servers, storage, and per-use
licenses. And that's not even considering the cost of training, maintenance, and
tweaking prompts or configuration.
Conclusion? Using AI only makes sense when it solves something that can't be
solved otherwise, or when the return on investment justifies it. If it can be
automated with simple logic, all the better. It's cheaper, more sustainable.
Automate wisely: what we actually do at Novatium
At Novatium, we believe that automation is a concrete tool, not a passing fad.
Our approach always starts with a key question:
What part of your operation is taking up human time on repetitive, non-value-added
tasks?
That's where automation is important.
With a clear methodology, we work like this:
1. Process diagnosis → We identify duplicate tasks, bottlenecks, and points
of failure.
2. Practical redesign → We eliminate unnecessary steps and prepare the
process for scaling.
3. Realistic automation → we use the most appropriate tool (scripts,
integrations, RPA, APIs, bots, or AI when it really adds up).
Real-life cases: automations without AI, but with high impact
These are some implementations we did with our clients, without promising miracles,
but with visible results :
- Automatic reset of internal passwords, freeing the help desk from more than
300 tickets per month.
- Automated bank reconciliations between Excel and ERP, speeding up accounting
closing.
- Expense and payment approval workflows, connecting email, SharePoint, and
financial systems.
- Automatic validation of CBU and CUIT for payments, avoiding rejections and
reprocessing.
- Scheduled accounting tasks, such as provisions or amortizations, executed
with simple but effective logic.
And when do we use AI?
We apply AI when it really generates value, for example:
- Natural language processing (emails, legal texts, chats)
- Extracting information from PDFs or scanned documents
- Automatic classification of large volumes of data
- Predictive or anomaly detection models in operations
We use it when necessary. Not to justify a tool, but to solve a real problem.
At Novatium, we don't sell magic bullets.
We help you solve processes with technology, common sense, and experience.
Sometimes, well-thought-out automation is worth more than a promise with a fancy name
Complete your application on this form