Ludwig Zuluaga is the founder and CEO of Appropria, which helps organizations automate document-intensive business processes using AI, workflow automation and enterprise content management. He tells Laurent Cochet what he learned from starting with a technology instead of a problem, why cultural change is the hardest part of enterprise software, and how AI agents are changing the way his company onboards and serves clients.
Starting from the problem, not the trend
Zuluaga says his first steps with the company went in the wrong direction. He was enthusiastic about open source software and how to bring it to enterprise, education and the public sector, and he believed open source technologies would take over the world. Starting from a technology trend, he says, meant learning about product market fit the hard way.
His training is in business administration and international business, not engineering, but he started programming in high school and later worked as a project manager, delivery manager and product leader. After fifteen years in the field, he can hold technical conversations about security, infrastructure and networking while also understanding the business. His warning to founders: simply saying AI is not enough without understanding the problem and the niche you serve.
Organizing documents scattered across systems
Appropria works in three areas: document management, workflow automation and agentic AI. It builds on top of the systems companies already run, such as ERPs and CRMs. In his experience, those systems and human resources software produce many documents that are scattered and unmanaged, so finding everything related to one provider can mean searching three or four systems.
The company centralizes those documents and adds metadata, such as expiration date, owner, department and how long a file must be kept, so files can be searched and reported on like a database. Collecting documents for an audit, he says, can take weeks without this structure and a few minutes with it. Approval workflows that once happened over email or WhatsApp are set up so they can be tracked, with a history of each document.
Because back office processes such as invoicing and accounts payable look much the same across industries, the company adapts its system to each client. It serves logistics, services, manufacturing and the aviation industry.
Why enterprise software meets resistance
The biggest challenge, in his view, is cultural change. The attitude that a process already works by email or WhatsApp is, as he puts it, the enemy of change. Structuring a process for long-term tracking and auditing often means giving employees more steps than before, and employees may dislike the new system as a result.
He argues AI is reversing that dynamic. Instead of adding work, agentic AI removes manual input and manual processes, so a system that once added tasks for users now takes them away.
From manual input to AI that understands context
Automating data entry used to require expensive OCR projects that were hard to implement. Now, he says, AI understands context and can extract data accurately and cheaply. One company expected two years of manual digitization; his team proposed automating it in a few months.
AI also handles work that once seemed to need people, such as cross-checking documents against complex business rules. He describes ingestion pipelines with more than two hundred AI-driven business rules that run in seconds, and clients whose document verification went from two weeks to three to five minutes. Beyond retrieving information, he sees agentic AI as a way to help people think better about their business.
Onboarding by group chat and growing abroad
Instead of starting with weeks of discovery meetings, Appropria adds one of its AI agents to a group chat with the client. Leaders answer its questions by voice message or document, name other leaders, and the agent interviews them too. One discovery covered about 15 people in two days. The company is preparing agents to run inside chat tools such as Google Chat, Microsoft Teams and WhatsApp.
Over the next 12 to 18 months, Zuluaga wants Appropria to lead in agentic AI for the enterprise. The Colombian company operates fully online and, by his count, has worked in around 10 countries, working with Canadian, US and Spanish companies. It is opening an office in Taiwan as a base for Asia, and it relies on technology partners such as integrators.
With unlimited resources, he would scale the agentic AI system, a field he says became clear to him when an open source agent project gained worldwide attention. His advice to founders is to build with users and get work into production, even in a small beta. The leadership habit he values most is inspiring the team to dream big.
Watch the full conversation
Watch the full conversation above. Ludwig Zuluaga says he can be found mainly on LinkedIn.
AI InterConnect hosts conversations with founders, investors and operators building with AI. Explore more episodes on the AI InterConnect podcast page.