Ramp Stack launches agentic close for accounting

Automation is moving from assistance to execution inside accounting firms. Ramp's new Stack platform signals a shift where AI agents can run the monthly close end to end, with auditability built in. That changes how work gets done and how firms price it. For firm leaders, this is not just another tool. It challenges the labor model behind core revenue. At the same time, risks are expanding at the identity layer, large firms are acquiring implementation capability, and platforms are opening to direct agent access. The common thread is clear: control is shifting toward systems that execute, not just advise. Also covered: a Microsoft 365 mobile security concern, Grant Thornton's acquisition strategy, Morgan Stanley opening platforms to AI agents, and key signals from GitLab, Coralogix, Google, Meta, and Uber. Learn more at https://crestvale.io

Transcript

SPEAKER_00 0:00

Your monthly close is about to stop being a team activity. And if that happens, the way your firm makes money changes with it. This is the Crestvale Newsroom Daily Podcast. Ramp just made its most aggressive move yet into accounting. They launch something called Stack. And this is not another assistant sitting next to your team. These are agents that actually run the close. That distinction matters. Most tools help draft. They suggest. They summarize. Stack executes. It pulls data, builds schedules, drafts entries, posts them, reconciles accounts, and logs everything back to source. End to end. With human checkpoints where you want them. If it works as described, the monthly close stops being a coordinated effort across staff and becomes a review layer sitting on top of automated execution. That is a very different operating model. What stands out here is not just automation, it is standardization. RAMP is pushing firms to encode their processes into what they call skills. That means your SOPs become structured workflows that run the same way every time, across every client. No variation, no dependency on who is assigned. That directly attacks one of the biggest inefficiencies in accounting. The gap between how different people do the same work. Now add auditability. Every action ties back to source data. Work papers are formula-based. Nothing posts without approval. That is not optional. RAMP clearly understands that trust is the barrier, not capability. So they are building defensibility into the system from the start. Here is why this matters. This is a credible attempt to compress the core of your revenue engine. If close work becomes faster, more consistent, and less labor intensive, the pressure on hourly billing shows up quickly. And if your firm is not the one delivering that efficiency, someone else will. This is not about whether automation is coming. It is about who controls it and how fast it reshapes pricing. Now, a very different kind of risk, but one that deserves attention. A reported issue in Microsoft 365 Android apps could allow other apps on the same device to access account tokens through an exposed debug setting. That is not a small bug. Those tokens can grant access without a password, which means many standard controls do not apply. This sits in the identity layer, and that is where the real damage happens. The bigger concern is how this happened. A leftover debug configuration. That is a preventable mistake, not a sophisticated attack. And it lives on mobile devices, where most firms have the least visibility and the weakest controls. If your team uses Microsoft 365 on personal or lightly managed Android devices, this is a blind spot. Closing it likely means tightening device policies, limiting access, or moving to managed environments. Because once identity is exposed, everything behind it is exposed. Meanwhile, Grant Thornton is making a clear bet on where firms compete next. They are acquiring MCA Connect, a firm with deep Microsoft capabilities in manufacturing and supply chain systems. This is not about adding more advisors. It is about owning implementation. Clients are not just asking what to do anymore. They want firms to design, build, and run the systems that execute that strategy, especially in industries like manufacturing, where AI and automation are already being deployed inside core operations. Grant Thornton is moving closer to that execution layer. And deals like this show a pattern. Firms with real delivery capability in enterprise platforms are getting acquired. Firms without it are getting boxed out. If your firm still stops at recommendations, the gap is widening. And then there is Morgan Stanley. They are opening their stock plan platforms to external AI agents using model context protocol. That means agents, not humans, can interact directly with systems like ShareWorks and Equity Edge. Pull data, execute tasks, move workflows forward, without logging into a user interface. This is a shift in how software gets used. The interface is no longer the application. It is the agent sitting on top of it. Morgan Stanley is betting that controlling data and workflows matters more than controlling the user experience. And they are moving early by allowing external agents, not just internal ones. For firms, this points to a clear direction. Your clients will expect your services to plug into their systems the same way. Agent to system, not human to software. If your workflows still depend on people logging into client tools and moving data manually, that model is already under pressure. Here is what else is worth knowing today. GitLab is cutting 14% of its staff while rebuilding around agent-scale workloads. The constraint is shifting from people to infrastructure. Coralogics raised $200 million focused on observability for AI systems. If you cannot see what your agents are doing, you cannot manage delivery. Google is rolling out AI-powered fake call detection. Voice is becoming a real attack surface, not just email. Meta is pushing business agents into messaging platforms. Client interaction is moving into chat, and response speed will start to matter more. Uber has capped internal AI spend after overruns. The shift from experimentation to cost control is already underway. Before we close out, here is a quick look at where markets landed. Equities closed lower in the previous session, with both SPY and QQQ pulling back. The 10-year yield moved higher, continuing its recent upward trend. In commodities and digital assets, Bitcoin and gold both declined, while oil pushed higher. Here is the takeaway. If your core workflows cannot be executed by agents with auditability, your pricing model is already at risk. Tomorrow we are watching how quickly firms move from AI pilots into full workflow automation, and which platforms actually get adopted inside real client work. If this was useful, follow the Crestvale Newsroom Daily Podcast so you don't miss it. Thanks for listening.