By Matheus Dias (Co-founder of Luzid), in partnership with Bruno Dias (Head of AI Innovation) and Isabella Figueiredo (Journalist – Numen)
This article explains how our internal use of Luzid’s AI copilot has been so far and what we’ve learned by incorporating this technology into our consultants’ daily routines.
Enterprise projects often run into the same reality:
- an enormous amount of required documentation,
- manual and time‑consuming steps,
- high dependency on experienced consultants,
- and a growing volume of change requests due to a lack of initial clarity.
Luzid emerged with an ambitious proposal: to automate part of this “invisible,” yet essential work, using an AI agent called David.
David maps processes, generates documentation, structures blueprints, and supports tasks from pre‑sales all the way to go‑live. In some cases, the startup states that automation can reduce up to 90% of the time spent on complex artifacts.
This does not replace the consultant — but it removes manual work so they can focus on architecture, make better decisions, and accelerate delivery.
Numen’s journey with Luzid: from curiosity to recurring usage
Our story with the copilot began by using David to speed up and improve the quality of BPDs (Business Process Documentation) in September.
Instead of starting with a “blank document,” our consultants began working from structured content, speeding up discussions and increasing clarity during the solution‑design phase.
But this beginning also exposed the biggest challenge: encouraging day‑to‑day adoption.
The main attention point in the first months was ensuring that David became a natural part of consultants’ daily routines.
This “cultural resistance” is a pattern across the market, not something exclusive to Numen. That’s why we structured a three‑phase plan — which ended up reflecting Luzid’s own evolution as a platform.
Phase 1 — September: The kickoff with BPD
The first consultants who tested it quickly saw value in the speed of generating artifacts. However, adoption was low and inconsistent. It was clear: we needed to reduce friction and make the tool more accessible.
Phase 2 — November: prompt customization and increased adoption
Luzid launched a new version of the platform allowing prompt customization — a major turning point.
With this customization, we were able to adapt David to Numen’s reality, to the style of our artifacts, and to our project language. This brought the copilot closer to the consultant, who now felt that the tool “spoke the same language.”
As a result, adoption grew consistently.
Phase 3 — January: Numen’s data integrated and usage tripled
The most significant shift came at the beginning of 2026.
By bringing our own data into the platform, consultants began using David for even simpler and more natural tasks — such as daily documentation, quick drafts, and support in understanding processes.
This change in behavior caused the number of users to triple.
With that, usage stopped being a “pilot project” and became routine.
Today, more than 10 projects use the copilot to some extent, totaling 50 distinct users who have gained enough familiarity to:
- generate deliverables faster,
- connect content directly to SAP Signavio,
- work with segmented information from the start,
- reduce rework time,
- decrease change requests.
In other words: the tool is no longer an “accessory” — it has begun influencing overall project quality.
Our focus now: removing friction and expanding value
At this stage of the journey, our efforts are on reducing adoption friction, ensuring that:
- all relevant content is available and organized within the platform;
- David is naturally triggered by consultants and becomes an integral part of the standard delivery flow;
- project testing is significantly accelerated through David;
- and risks are monitored holistically, from the initial assessment to go‑live.
As usage grows, we are also able to test new workflows, suggest improvements, and co‑create strategies with Luzid to generate even more value for our clients.
We are learning — and evolving — fast.
What does this mean for the future?
Our journey with Luzid’s copilot confirms that applying AI in practice does not follow a single path — it is a process of testing, adjusting, and discovering. Just as Luzid itself evolves and pivots its solutions, Numen is also learning continuously, experimenting with different approaches, and understanding where technology truly generates impact.
And we continue moving forward with the same purpose — transforming this constant learning into real value for Numen’s entire client portfolio.
Matheus Dias is co-founder and COO of Luzid, a startup offering AI solutions to accelerate corporate software implementations. During his time at Stanford, he followed the rise of LLMs in 2022 and became a founding member of Orby AI, collaborating with companies such as Google, Uber, and Airbnb.
