Skip to content 
Every client engagement starts the same way: we learn as much as we can about the client so we can support them in reaching their goals.
This means understanding the client in terms of:
- Objectives
- Pain points, both business and technical
- Team structure
- What their platform looks like today
- Where they want to be*
*Sometimes clients come to us when they are still unsure of the end goal. We help them figure that out too.
Clients differ so much from one another that no single solution or architecture fits every case. Success here comes from experience combined with an approach tailored to the specifics of each client.
The importance of an individual approach
For every consulting engagement, we map priorities and weigh the pros and cons of different paths.
Our role is to guide the client towards their future state with their best interest in mind, always making sure their team feels confident using the technology we build.
Real success for a data and AI company like Infinite Lambda means working with a client for a defined period, then handing over a solution they can scale and maintain without difficulty. That is the path we aim for, and one we have tested across hundreds of projects.
On Google Cloud Platform
With that in mind, let me share an overview of two solutions we built using Google Cloud Platform (GCP) as part of the architecture.
These cases show how expertise in these ecosystems, combined with a solid approach, matters for building a scalable platform with robust processes in place.
GCP keeps proving it is very much in the game as a modern data stack, releasing strong features and improving its offering based on client needs, with the AI era in mind.
It offers genuinely good tools, but at enterprise scale, no single tool solves problems on its own. To truly empower a company, we need strong architecture and process alongside it.
In this article, I will look at two projects where the problem really came down to trust. The first was about trusting that the data itself was correct and easy to find. The second was about trusting what happened the moment that data changed.
Building a GCP data platform with strong governance
The first case is about putting in place a data platform with data governance built in from the beginning.
The project used BigQuery and Knowledge Catalog, and the insights I am about to share come from my own involvement as project lead.
If you work in enterprise data, you will know this combination well: nobody knows where to find the correct data, there are multiple numbers and reports for the same metric, business teams cannot find answers fast enough to support decision makers, and if you are lucky enough to have any documentation at all, it is poor and outdated.
This is the challenge we help companies resolve, and it is the one I want to tell you a story about.
When the problem is systemic, everyone knows something is wrong, yet it seems impossible to put a figure on the time and budget lost to inefficient systems.
In our case, the client was already on GCP, so their employees were used to the platform. Our job was to modernise the environment, delivering an efficient solution within that ecosystem that would bridge the gap between business and technology.
Leveraging Knowledge Catalog
Knowledge Catalog is a universal context engine that helps agents carry out complex tasks accurately. It is a powerful tool for the agentic enterprise era, and it proved key to this project.
Formerly known as Dataplex, Knowledge Catalog marks a shift from a traditional, passive data framework to an active, AI centric governance layer.
Knowledge Catalog aggregates data across the whole platform and keeps learning continuously, rather than sitting still as a static governance tool where the business context is always out of date.
As with most enterprises, a lot of business knowledge had never been documented, or lived scattered across multiple systems.
We decided to bring Knowledge Catalog into the architecture from day one, and it made all the difference.
Instead of waiting for someone to update the documentation manually, which turns that person into a bottleneck, Knowledge Catalog harvests the metadata itself, continuously. It uses Gemini to extract business meaning from raw data, keeping that context current without needing a human curator in the loop.
Shared knowledge
Knowledge Catalog was built for AI agents to query before they act. This prevents hallucinations, and access control carries through as well, so an agent can only see the data that the specific user querying it would be allowed to see.
Semantic definitions are also shared between humans and agents, so the two never disagree on what a metric actually means.
Practitioners benefit too: the tool supports the business without creating extra work for anyone.
If shared knowledge and accuracy is something you are struggling with, Knowledge Catalog is certainly worth a look.
Smooth deployments
The second solution comes from a project led by Analytics Engineer Othmane Benbella. It uses dbt and GCP to solve blue-green deployment and ensure safe production changes.
This solution was originally built for a particular client, but the approach has since been reapplied across multiple projects, since it works for any GCP and dbt setup, helping make deployments safer.
The client had been facing real operational challenges around deployments. Releases would take more than two days, and during that window, users could not query the affected tables.
The result was at least two days where the business could not fully trust its own numbers, with important decisions getting delayed and users losing trust in the data.
The rollback process was manual and error prone, which made the team's own fallback plan a risk in itself.
This becomes a vicious circle: riskier releases lead teams to deploy less often, which means bundling more changes together, which then increases everyone's anxiety around each deployment.
To break this cycle, we created a dbt package that integrates with BigQuery. It solves these problems by using BigQuery's own native capabilities, making zero-downtime deployment possible without extra infrastructure or cost.
The solution ticks the technical boxes, but more importantly, it has made the team feel safer about deploying changes ever since launch.
This is a good example of what consulting is really about. The solution was not part of the original scope. As consultants, though, it is our job to notice a problem, address it proactively, and build something effective that supports everyone on the client side.
It is this kind of approach, one that tackles real user challenges, that brings lasting success.
If you are interested in this solution, stay tuned for a separate article that will go into more detail and look at the impact it has had across client organisations.
The consultant mindset
These two solutions show what a successful consultant mindset looks like: understanding the client's pain points and taking a proactive approach to solving them.
Infinite Lambda clients vary in size, team structure, priorities, business and technical pain points, and culture. Our approach, therefore, is always to guide the client towards a solution that best fits their business reality.
This principle matters more than any platform or technology, since it is rooted in judgment and best practice.
GCP in the modern stack
In these two cases, the best solutions happen to live on GCP, which holds a significant place in the modern data ecosystem.
GCP keeps evolving to meet growing enterprise demands, speeding up people's work and enabling agentic capabilities.
If the challenges above sound familiar, whether it is a data catalogue nobody trusts or a deployment process everyone dreads, reach out. We would love to talk through how the modern data stack can help you address them.
