Most finance tools are built by people who understand finance. Runway feels like it was built by someone who found finance confusing and decided that was a problem worth fixing. Siqi Chen, the CEO, has said publicly that financial expertise always felt like something for "real business people," and that frustration with spreadsheets was what motivated the company. That is an unusual founder origin for a financial modeling tool, and it shows in the product.
The result is a platform that startup founders and operators actually want to use. Customers include Superhuman, Stability AI, and Sandbox VR. The $33.5M in funding includes Initialized Capital leading the Series A, Garry Tan joining the board, and angels including Elad Gil and the founders of Dropbox, Notion, and Product Hunt. That is a concentrated network of people who understand product quality at a high level.
The Market
A category being rebuilt from scratch by AI.
The FP&A software market was $3.9B in 2024 and is on track for $9.7B by 2032, growing at 16.4% annually. The market is growing not because companies are buying more of the same thing but because the thing is changing. Gartner projects AI will automate 80% of routine FP&A tasks. What that means in practice is that the planning layer, which has historically been a collection of Excel exports, email attachments, and quarterly review decks, is being replaced by software that can actually respond to changes in real time.
The incumbent tools, Anaplan, Workday Adaptive Planning, Planful, were built when the hardest problem was getting the numbers in the same place. The new problem is keeping them current, collaborative, and interpretable without a finance degree. That is a different product problem.
The Landscape
Where the existing tools stop short.
live sync
editing
forecasting
UX
● Purpose-built◐ Partial / legacy○ Not a focus
Anaplan and Workday Adaptive Planning were built for large enterprise finance teams with dedicated administrators. They are powerful but opaque, and they require significant implementation work. Most growing companies end up on Excel because the enterprise tools are too heavy and there is nothing useful in between. That is the gap Runway occupies.
The Engineering
The hardest part has nothing to do with the interface.
Financial models are fundamentally relational and dynamic. A change upstream, a new headcount plan or a revised revenue assumption, should cascade correctly through every dependent calculation in real time, across a shared collaborative workspace, without breaking anything. Building that with spreadsheet-level flexibility but software-level reliability is the core technical problem. Spreadsheets are forgiving. Shared cloud applications with real-time sync and multiple concurrent editors are not.
The integration layer is also underappreciated. Pulling live data from QuickBooks, NetSuite, Stripe, and payroll systems, reconciling it into a coherent model, and keeping it current without manual exports is a backend problem that requires real systems thinking. I have built versions of this problem: at Amazon, I built a consistency tool reconciling pricing data across S3 and DynamoDB. The cascading-update problem is non-trivial and the failure modes are subtle.
The hardest part is not the interface. It is keeping the numbers consistent when five people are editing the same model at once.
The Frontier
From reactive reporting to proactive intelligence.
Runway launched what they call an Ambient Intelligence Platform for Finance Teams in September 2024. The idea is that the model learns from how a team makes decisions, surfaces anomalies and risks before the monthly review, and gradually becomes less of a tool the CFO opens and more of a system that runs in the background. That is a meaningful shift in what FP&A software is supposed to do.
Whether that bet works depends on whether the AI layer can actually learn the right things from financial data that is noisy, retrospective, and full of one-time events. Finance data is structurally different from the kind of data AI systems are typically trained on. The teams that figure that out will have built something genuinely hard to replicate.