Workshop
Practical AI for Finance
Aerium Team
Workshop on how AI is transforming the finance function across manufacturers
What we're seeing
How AI is evolving in finance
As a function, finance has been at the frontier of AI transformation because it is directly tied to the business' P&L. Over the past 36 months, the scope of work for AI has broadened from being a glorified Google search to automating reports, managing expenses, and handling quoting across multiple business units.
Where things are going
We are seeing two main trends across our deployments and across the broader AI industry: The best intelligence is built, not bought and agents are beginning to take on tasks that span multiple days to weeks.
1. The best intelligence is built, not bought
Companies are building in-house teams or partnering with external firms to train proprietary AI models on their internal data. Proprietary models allow for customers to maintain complete control over what happens with their data and allows them to have models that perform better than ChatGPT or Claude.
2. Agent systems are beginning to span days and weeks across multiple parties
AI agents are beginning to run over long, multi-day time horizons completing entire bodies of work across multiple stakeholders, only surfacing for high level/admin approval, questions, or when tasks are completed.
Together, proprietary models and long-running agent harnesses allow companies to automate large portions of work that were previously impossible.
Where AI can help you
Getting started
The best way to start using, experimenting with, and benefitting from AI is to just start using cowork. Let's see what it can do!
How it's being deployed at an industrial manufacturer
At one industrial manufacturer, AI is being deployed to accelerate quoting and deliver 100% visibility across the backlog. Custom models were trained on historical BOMs and cost details to predict unit costs for any configuration without requiring engineering team input.
Additionally, the FP&A team is leveraging claude cowork to automate many other parts of reporting and analysis.
Identifying finance workflows where AI can deliver ROI
AI can be a very powerful tool, and it is best used when it is tied to a measurable business objective such as for quoting, inside sales or improving forecasting.
Working session
Getting started with Cowork
In groups of 4-5, discuss how you think AI could be used at 3 levels:
- Within your day-to-day role
- Within your team
- Across the company as a whole
Some examples to start the conversation:
- Run an analysis to compare the performance across our plants over the past 3 months given this excel file
- Send me a report every morning about the top 3 priorities based on yesterdays work and email traffic
- Pricing and sensitivity analysis: Conduct price sensitivity analysis by analyzing historical and current market sentiments
Lets try it!
Open cowork and explain the task you do in your day-to-day role and try to create a skill to assist you in completing it faster. If you're unsure where to start, try explaining your work to copilot and see what it recommends you try first!
Questions?
Appendix
What is a token?
Tokens are the fastest growing spend category
With the explosion of AI usage, tokens have become the fastest growing spend category1. Managing tokens can be challenging at the organizational level, but at the individual level, one can route tokens.
AI safety and ethics
Our stance on the ethics of AI is that it should be used as a tool to enhance human potential rather than a way to replace human labour. For example, when we build finance automation workflows, our goal is to allow a team of 3 to operate like a team of 10. Additionally, safety is a function of the implementation, and by owning the underlying models, data, and guardrails, reliability becomes a feature.