I fully agree with the data warehouse/one source of truth we can plug AI on.
For the rest I feel that the use of Claude still requires a lot of checking for monthly closing that a strong well thought process and excel avoids. For adhoc analysis, however.. a game changer
This was a great read. I'm currently working on getting the Level 2 in place. Any tips on how/what to use to build the semantic layer? I'm currently testing using BigQuery as that's the warehouse where our non-financial data sits
thanks for the kind words, Aziz! i had the same question -- so i interviewed Julia King, Gusto's head of data, as a followup to the Gusto / Jeff Cobourn interview! coming soon :)
and to answer your question re: which semantic layer to use -- she recommended https://cube.dev/ which is what they use.
I disagree that this is what the evolution of AI in FP&A is going to look like for a few reasons.
1. Teams move across levels - someone can go from Level 1 to 3 or 2 to 4 directly. AI now allows for teams to leapfrog levels based on what matters to each team.
2. The article reduces the work FP&A teams do to spreadsheet crunchers and data consolidation etc which isn’t the case. Team FP&A teams also spend a lot of time analysing the data and working with business finance (where these teams exist) or directly working with the business to move actual underlying business outcomes.
3. Where are the MBRs (Monthly Business Reviews)? A key element of an FP&A team is to give inputs during the Exec Staff meeting or running the MBRs where we hold the P&L owners accountable, show them numbers and discuss a path forward.
This article assumes that all the monthly bottom up re forecasting happens without inputs from the business teams and only based on actual after the close.
So for all FP&A leaders who read this and wonder where is the actual work your teams do, you’re not alone.
Thanks for clarifying that Julian and getting back to me. I appreciate it. Even for the MBRs and the Strategic work, AI can help make the MBR decks or help prep for calls with stakeholders, prepare customised dashboards specific to individuals, to name a few.
i brainstorm a little bit with AI, and i use AI transcription tools to record interviews, but all my posts are 100% human writing and 100% human editing.
another excellent piece on AI + Finance. would love to see who/when/how a team gets to level 4.
thanks Jimmy! we're building something that will unlock level 4 :) watch this space!
I fully agree with the data warehouse/one source of truth we can plug AI on.
For the rest I feel that the use of Claude still requires a lot of checking for monthly closing that a strong well thought process and excel avoids. For adhoc analysis, however.. a game changer
100%. how are you handling the DW / semantic layer on your side?
Still SQL server but looking into Microsoft Fabric
This was a great read. I'm currently working on getting the Level 2 in place. Any tips on how/what to use to build the semantic layer? I'm currently testing using BigQuery as that's the warehouse where our non-financial data sits
thanks for the kind words, Aziz! i had the same question -- so i interviewed Julia King, Gusto's head of data, as a followup to the Gusto / Jeff Cobourn interview! coming soon :)
and to answer your question re: which semantic layer to use -- she recommended https://cube.dev/ which is what they use.
I disagree that this is what the evolution of AI in FP&A is going to look like for a few reasons.
1. Teams move across levels - someone can go from Level 1 to 3 or 2 to 4 directly. AI now allows for teams to leapfrog levels based on what matters to each team.
2. The article reduces the work FP&A teams do to spreadsheet crunchers and data consolidation etc which isn’t the case. Team FP&A teams also spend a lot of time analysing the data and working with business finance (where these teams exist) or directly working with the business to move actual underlying business outcomes.
3. Where are the MBRs (Monthly Business Reviews)? A key element of an FP&A team is to give inputs during the Exec Staff meeting or running the MBRs where we hold the P&L owners accountable, show them numbers and discuss a path forward.
This article assumes that all the monthly bottom up re forecasting happens without inputs from the business teams and only based on actual after the close.
So for all FP&A leaders who read this and wonder where is the actual work your teams do, you’re not alone.
hi Naga, thanks for subscribing. this article is focused on the areas in which i’m seeing AI improve FP&A productivity.
MBRs and strategic work are still jobs best done by humans, which is why they aren’t covered here.
Thanks for clarifying that Julian and getting back to me. I appreciate it. Even for the MBRs and the Strategic work, AI can help make the MBR decks or help prep for calls with stakeholders, prepare customised dashboards specific to individuals, to name a few.
Expected when the article is 80% written by AI 🙄
i brainstorm a little bit with AI, and i use AI transcription tools to record interviews, but all my posts are 100% human writing and 100% human editing.