Empowering Devs, DevOps and Finance Teams with GenAI-driven Cloud Cost Intelligence

I was talking with a CFO about multiple ways to use GenAI in his world, and one of the ideas I suggested to him was: "Why don't you start with the Cloud Costs data?"

Below is an example of what could look like.

For reference, this article was created in about 1h using:

• ChatGPT 4o voice mode: to transcribe my ideas and provide some initial ideas consolidation
• Claude 3.5: to create a big 'brain dump' document, which captures most of the ideas and concepts in a coherent document
• Claude 3.5: to consolidated all those ideas in the first draft of the article (which I then copy and pasted here on LinkedIn and made multiple changes before publishing)

Coming from the cybersecurity world, I learned to focus on the wealth of information hidden in cloud billing data

While teams usually focus over logs and access patterns, there's another goldmine of insights sitting in plain sight: the detailed cloud bill.

It's not just about costs - it's a comprehensive record of what's actually happening in your cloud environment.

Here is an interesting approach that turns cloud billing data into a powerful tool for business alignment and technical optimisation.

The key? Starting with the finance team.

Why Finance Teams Hold the Key

Here's something that might surprise you: finance teams are uniquely positioned to drive cloud optimisation.

They have:

• Direct access to detailed cloud billing data
• Visibility into organisational structure and project funding
• Understanding of business objectives and ROI requirements
• The ability to connect technical spend with business outcomes

... and ultimately control the budgets :)

But here's the challenge, most solutions (and unfortunately a good number of finance teams) focus solely on cost reduction, missing the bigger opportunity to drive value optimisation and business alignment.

A Different Technical Approach

The technical solutions and architecture that I've been developing at The Cyber Boardroom, take an intentionally simpler but powerful approach.

Instead of building complex databases and real-time processing systems, I tend to follow this pattern:

• collect the data in raw format
• store it in file systems (usually cloud based)
• apply multiple transformations (some using GenAI) whose results are also stored in the file system
• all exposed/driven by APIs
• all created/deployed using fully automated CI/CD Pipelines
• all running on serverless environments (or disposable Kubernetes environments)

I create the above via collaborative sessions with GenAI models, where we focus (and debate) on: architecture design, schema creation and some of the transformation code

In this specific use case ("Empowering Devs, DevOps and Finance Teams with GenAI-driven Cloud Cost Intelligence") , we start with collecting the billing data and storing it in its raw form.

Here's why this matters:

• Cloud providers already give us detailed billing data (via APIs)
• We don't need real-time processing (billing data itself is delayed)
• Simple file storage is more efficient than maintaining databases
• It creates a clean historical record for analysis

The workflow looks like this:

1. Collect billing data 5-6 times daily (optimised for cost and data availability)
2. Store raw data in cloud storage (not databases)
3. Transform data through clear, documented steps
4. Build relationship graphs connecting resources to projects

The Environment Management Revolution

The most exciting aspects of this approach are the changes to how Developers/DevOps/Testers operate, and how accountability is now owned by the correct business/technical stakeholders.

The principle is simple: you should only have resources active when you need them, and only at the scale you need them.

This sounds obvious, but it's surprisingly hard to implement. Cloud providers make it easy to spin things up but don't really optimise for spinning them down.

We're changing this with:

• Automated environment creation and teardown
• Clean environment deployments
• Multi-region and multi-cloud capabilities
• LocalStack for local development

I've been using LocalStack extensively in this context, and it's a game-changer. It allows developers to work with cloud services locally, making the development process faster and more efficient.

Beyond Technical Metrics

Here's where it gets really interesting. For this to work we will need much more than just collecting billing data, we will need to build a comprehensive graph that includes:

• Cloud assets and configurations, for example from existing Terraform, Helm charts, or CloudFormation templates/scripts
• Project lifecycle stages (development, testing, production), for example from Issue Tracking systems like: Jira, Asana, Notion, GitHub, GitLab, etc...
• Business objectives and strategy alignment, most likely from spreadsheets
• Risk register entries and compliance requirements, from risk mappings tools (or spreadsheets)
• Organisational structure and responsibilities (i.e. the business org-chart)

This will create powerful feedback loops that will helps and empower teams/stakeholders to make better decisions. For example, a development team can see not just their current cloud spend, but how it relates to:

• Project outcomes and business objectives
• Risk reduction goals
• Compliance requirements
• Overall business strategy

The GenAI Advantage

Generic AI plays a crucial role here, but not where you might expect. While it can help with data transformation, its real power comes in creating personalised views of this information for different stakeholders.

For example a CFO needs different insights than a development team lead.

GenAI helps us transform technical data into relevant insights for each persona, taking into account:

• Cultural context
• Technical expertise
• Business knowledge
• Role responsibilities

This 'mapping to persona' is absolutely critical, and an area that I've spent quite a lot of time researching and figuring out how to do it at the Cyber Boardroom

Real-World Impact

What makes this approach powerful is that it's not about cost reduction (though that often happens as a side effect). Instead, it's about empowering teams to make better decisions.

I've seen teams transform their development practices when they have this kind of visibility. They:

• Build more efficient architectures
• Develop better testing practices
• Create reusable patterns
• Make more informed technical decisions

A Business Model for a startup?

Everything described here could be provided as a service, specially if it is able to run entirely within a client's environment, where:

• Sensitive data stays within the client environment
• Integration is simpler and more secure
• Deployment is clean and containerised
• Scaling becomes straightforward

This is possible via deploying the entire solution in a clean cloud account, making it perfect for client adoption while maintaining security and privacy.

Making better decisions

The combination of finance team insights, technical optimisation, and business alignment creates a powerful framework for better cloud resource management, and ultimately allow for better business decisions, better ROI and faster delivery of value to the users or customers.

Until GenAI, I could never see how this could scale, now I can :)