# Figma finance team stops filing tickets, builds _Figma CFO Preer Milwani told a16z his finance team now builds its own AI tools, shifting from SQL and Excel to Codex and Figma Make._ **Published:** 2026-11-05 **Source:** https://www.startuphub.ai/ai-news/artificial-intelligence/2026/figma-finance-team-stops-filing-tickets-builds --- [a16z](https://www.youtube.com/watch?v=9iAQdKxyiQc) captured what changed inside [Figma](/startups/figma) when CFO Preer Milwani told host CMA Amble and partner Yvon Macarov his finance team started building its own software. The shift is practical. Instead of living in 400 to 500 line SQL queries and Excel macros to orchestrate data from Netsuite, Salesforce, [Stripe](https://www.startuphub.ai/ai-news/artificial-intelligence/2026/stripe-s-ai-strategy-build-everything-faster), Non, Zip and RAMP, staff now prompt [Codex](https://www.startuphub.ai/ai-news/artificial-intelligence/2026/openai-s-40b-revenue-run-rate-fuels-ipo-hopes), ChatGPT agents and Figma Make to stitch workflows together. Figma Make, per Anima, builds a working prototype or web app that you can preview, refine in chat, edit in code and publish, which is how the team moved from static models to runnable internal tools. Reconciliation flipped from dread to daily habit. Milwani said reconciliation was a very negative word associated with weeks and months of chasing, and now it is something you do or even a positive word because you can solve it much easier and quicker with modern tools and with AI, according to a16z. He put the data plumbing at 95% of the challenge, and said speed has compressed weeks of analysis into an hour, letting the team catch anomalies early and save hundreds or thousands of dollars. That daily access is why finance is expected to act like owners, not ticket filers. The examples are specific. Kajal on credit and collections rebuilt his workflow tied to multiple systems and now spends time driving down days receivable. Corp dev vibe codes products during diligence to test depth. Corporate finance fuzzy-matched T&E across three sources for the 27 planning cycle and auto-generated policy suggestions and visualizations. One growth-stage team Milwani cited spun up 40 or 50 agents in one to two months, but he estimated 90% of those tools are not used anymore. That waste is the tell. Milwani said the messy middle is here. Teams are building faster than a central systems function can productize them, which is why Figma just added Casper Sorenson from Coinbase as CIO yesterday to create a PM-like layer that hardens bottom-up builds so they survive beyond their creator. Audit-ready output for business-critical systems is still an extremely high bar that will require auditor conversations, and AI matching today is directionally correct, not definitive. Excel is not going anywhere. Milwani told a16z, "No. It's like my love language, dude," and argued the sheet is for building intuition the way Figma lets designers riff by hand. Models now help read those multi-tab monsters and surface sensitivity, while universities, he said, have reverted to fundamentals as syllabuses churn with each model drop. The hiring filter has shifted to intellectual curiosity, data fluency and structured communication, with a case study for VP finance that tests whether candidates understand what a model output actually means. The net effect is fewer handoffs and more judgment work. Finance is moving from reconciling data to running experiments, pricing studies and competitive benchmarks, often demoing builds in the company all-hands segment called Fig Nation. If complexity stays flat, Milwani said they probably would not need the same headcount a year from now. --- Original analysis from [startuphub.ai](https://www.startuphub.ai), the #1 AI startup directory. © StartupHub.ai. All rights reserved. You may not republish this article in full without a license. Search engines and AI research tools may crawl and summarize for reference. Bulk reproduction or model training on this content requires a license. See https://www.startuphub.ai/terms.