The transformation of the finance function has historically been structured around two complementary dynamics. On the one hand, the modernization of ERPs, responsible for unifying accounting, purchasing, treasury and reporting within an integrated base. On the other, the rise of FP&A type management platforms, which offer more granular forecasts, collaborative modeling and planning gradually freed from Excel.
In recent months, a third front line has been established, more technical, and often more decisive for the operational reality of the teams, because it is not primarily aimed at planning, but at execution. It does not necessarily replace ERP, but seeks to regain control of what makes it usable on a daily basis, namely transactional data, its consistency, and the ability to automate workflows in a heterogeneous environment.
The problem that no one really “owns”: broken finance
In business, finance rarely lives in a single system. Transactions are divided between ERP, files, business tools, data lakes, historical systems, and integrations built over time. This fragmentation imposes workaround routines. The teams spend a lot of time reconciling and reclassifying; a significant part of the effort consists less in analyzing than in reconstituting a common truth.
It’s not a glamorous subject, but it’s the condition for everything else. As long as the data is not stabilized, the AI remains a surface assistant, confined to text production, writing assistance, or summaries. In a closure and audit environment, the challenge lies elsewhere: reliability, traceability, reproducibility.
🚨 SMARTJOBS
- ECOLE POLYTECHNIQUE – Director/Deputy Director of International Relations (F/M)
- CLAROTY — Sales Development Representative
- CURE51 — Data Scientist (Internship)
- FRACTTAL — Account Manager (France)
- BRICKSAI — Founding Growth Manager
👉 Find all our offers on the DECODE MEDIA Jobboard
📩 Are you recruiting and want to strengthen your employer brand? Discover our partner offers
The “infrastructure layer”: the new object of desire
Stacks presents itself as an AI platform for accounting and corporate finance teams, with the idea of building a data layer connected directly to finance systems in order to produce a single, coherent and actionable financial view. This layer aims to standardize the data and make it immediately actionable for automation. We no longer talk about planning better, but about executing cleaner.
Why agentification attracts investors and sometimes worries teams
Stacks emphasizes “agents” capable of automating workflows across the finance stack. This is where the subject becomes strategic. In an accounting context, the agent has no right to approximation and must produce a result that is verifiable, documented and compatible with internal control and often audit requirements.
The traction of this segment is due to simple convergence. On the one hand, financial departments are under pressure, regarding closing deadlines, the quality of reporting and the ability to explain performance. On the other hand, the necessary technological building blocks are finally becoming available: more standardized integrations, more mature data infrastructure, models capable of processing context and growing demand for automation that goes beyond simple RPA.
The point of tension lies in governance. The company can accept a co-pilot who “suggests”. It is less tolerant of a system that “acts” without safeguards. This is what distinguishes product communication from the reality of adoption: the question is less whether the agent is spectacular than whether it is controllable.
The real topic: who will control the value chain above ERP
This battle for the infrastructure layer is not only technical, it concerns data ownership and orchestration capacity.
If a platform becomes the operational gateway to finance, it can gradually capture functions previously considered “native” to ERP or carried out in Excel. Conversely, large ERP vendors have obvious reasons to reabsorb these uses by integrating more AI and automation into their suites. In between, a gray area appears, the layer that connects to everything, unifies, explains and triggers actions.
Stacks as a symptom of a CFO switching from production to arbitration
The most interesting consequence is the redistribution of time. When finance spends fewer hours “putting together” the figures, it can discuss them differently. In practice, AI in the CFO’s office does not primarily replace people, but reduces certain manual actions, which implies a change in governance and supervision.
A trajectory funded in twelve months
Stacks, based in London, develops an AI platform for accounting and corporate finance teams, aiming to unify data scattered between ERP, spreadsheets and legacy systems and then automate workflows via agents. The company claims over thirty clients, including Future plc and Epidemic Sound. It was created by Albert Malikov, formerly of Uber and Plaid. Stacks announces a Series A of 19.6 million euros led by Lightspeed, with the participation of EQT Ventures, General Catalyst and S16VC, twelve months after a seed of 12 million dollars. In total, the company says it has raised $35 million.
