Batch processing is the target, not the incumbents themselves
Rillet's competitive pitch is narrower and more specific than 'better ERP.' The company is targeting batch-processing financial systems, the architectural pattern underlying most legacy ERP platforms, where financial data gets calculated and closed on a monthly or quarterly cadence rather than continuously. That architectural choice, batching rather than streaming, was a reasonable engineering tradeoff decades ago when compute was expensive and real-time processing across an entire general ledger was impractical at enterprise scale.
Rillet's bet is that the tradeoff no longer makes sense given current AI and compute economics, and that a system built from the ground up around continuous processing can offer CFOs something batch systems structurally cannot: financial visibility that updates in real time rather than only at the end of a close cycle that can take days or weeks to complete properly.
What the agents are actually doing
CEO Nicolas Kopp described the mechanism directly: 'Our agents help with that work 24/7 in the background, so CFOs have better financial information more quickly.' The specific tasks these agents handle fall into the category of ongoing accounting operations, reconciliations, categorization, and the kind of continuous bookkeeping work that traditionally accumulates and gets processed in a concentrated burst during close rather than being handled incrementally as transactions occur.
That shift from periodic batch updates to continuous daily visibility is the core product claim, and it maps directly onto a persistent CFO frustration: financial close has traditionally meant a multi-day scramble at the end of every month where finance teams work overtime to reconcile and report, often discovering issues only after the period they relate to has already closed. Continuous processing, if it works as described, moves that discovery earlier, when issues are still easier and cheaper to correct.
The four-to-five-year head start Botha is counting on
Roelof Botha's estimate that incumbent ERP vendors need four to five years to match Rillet's architecture is a specific, falsifiable claim worth tracking rather than dismissing as investor hype. Legacy ERP platforms were not designed around continuous processing from the start, and retrofitting that capability into decades-old codebases, transaction models and integration ecosystems is a fundamentally harder engineering problem than building it natively from day one, which is the advantage every well-funded challenger claims but few can substantiate with a credible technical argument.
Whether that estimate holds will depend heavily on how aggressively Oracle, SAP and Workday choose to respond. All three have been investing heavily in AI-native features layered onto existing platforms, and the real test is whether that layering approach can approximate Rillet's real-time architecture closely enough to blunt the competitive threat, or whether the underlying batch architecture creates a ceiling that incremental AI features cannot overcome regardless of investment.
Kopp's own admission about switching costs
Kopp's characterization of ERP replacement as 'like open-heart surgery' is an unusually candid admission from a founder trying to sell displacement of an incumbent system. ERP migrations are notorious across enterprise software for running over budget, over timeline, and carrying real operational risk during the transition window, and Kopp naming that risk directly rather than glossing over it suggests Rillet's sales motion is built around acknowledging the friction rather than pretending it away.
That candor is also a strategic bet: rather than positioning Rillet as a drop-in replacement for existing ERP, the more credible path to adoption is likely targeting growth-stage and mid-market companies that have not yet ossified around a legacy system, where the switching cost calculus favors starting fresh with a real-time architecture over migrating an established, deeply integrated incumbent deployment.
The labor shortage angle underneath the product pitch
Beyond speed, Rillet's pitch addresses a persistent and separate problem: accounting labor shortages that have made it harder for companies of all sizes to staff finance teams at the depth previous eras assumed was standard. Automation that lets a leaner finance team maintain the same or better level of financial visibility is a distinct value proposition from real-time reporting alone, and it may prove to be the more durable driver of adoption if the labor market for accounting talent stays tight.
For CFOs already struggling to fill open accounting roles, a platform promising to reduce the headcount required for routine bookkeeping and reconciliation work is a proposition that resonates independent of whether real-time close speed itself becomes a competitive necessity. The two value propositions reinforce each other, but they appeal to somewhat different buying triggers within the same finance organization.
What CIOs and CFOs evaluating this should watch
The article's own framing notes that deploying finance agents at this depth creates downstream security demands across infrastructure layers, a detail worth taking seriously given how sensitive financial system data is and how much autonomous agent access into that data represents a meaningfully larger attack surface than traditional batch-processed systems with more limited, scheduled data touchpoints.
For enterprise buyers evaluating Rillet or similar AI-native ERP challengers, the practical diligence checklist should include how agent actions are logged and attributed, what happens when an agent's automated categorization or reconciliation is wrong, and how quickly a human reviewer can catch and correct an error before it propagates through downstream reporting. A system processing continuously rather than in batches changes both the speed of errors surfacing and the speed of errors compounding if oversight does not scale at the same pace as automation.
