DualEntry Tasks Brings AI Agents Into Everyday Finance Work

DualEntry, the AI-native ERP built for modern finance teams, has launched Tasks, a new workspace that brings accounting work, approvals, exceptions, and AI agents into one place, according to the company's October 13 announcement distributed via Business Wire. The move represents another step in the startup's shift from an AI-assisted system of record toward what it calls an agentic ERP: software that does not just surface financial information but increasingly takes on finance work directly.

DualEntry Tasks is available now to the company's customers, with further details on the company's blog. The DualEntry Tasks launch lands one week after DualEntry announced a native two-way integration with Expensify, and it arrives as the race to put autonomous agents inside enterprise finance software heats up across the industry.

What DualEntry Tasks does

DualEntry Tasks reorganizes the finance team's day around a single live workspace. Instead of moving between different parts of the ERP to figure out what needs attention, the system surfaces work as it happens: agents, exceptions, alerts, and notifications can each automatically create tasks, giving teams a live view of what needs action, according to the announcement.

The agents inside DualEntry Tasks are designed for recurring finance work rather than one-off questions. According to the release, teams can use them for chasing overdue invoices, preparing finance updates, coding transactions, and coordinating approvals, with an audit trail recording what was done. Because Tasks is built directly into the ERP, it reduces the need for separate close trackers and manually maintained checklists; as work completes, the task board updates with it, so teams can see what has been resolved without maintaining a second record elsewhere.

The design keeps humans in the decision loop by default. According to co-founder Santiago Nestares, agents that have enough context to do a task on their own can handle it in the background and surface the result, pulling people in only where judgment or approval is required. The intended payoff, as described in the announcement, is a month-end close that advances continuously throughout the month, with less chasing, less clicking between screens, and more of the work already finished by month-end.

From AI-native ERP to agentic ERP

DualEntry Tasks is the latest expression of a strategy the New York company has pursued since its founding in 2024 by Nestares and Benedict Dohmen. The startup raised a $90 million Series A round co-led by Lightspeed Venture Partners and Khosla Ventures, with GV participating, valuing the company at approximately $415 million, according to the company's funding history as documented on Wikipedia, which cites Reuters reporting from October 2025.

The platform's calling card has been speed of adoption: its NextDay Migration capability claims to move a business's full historical financial data from legacy systems such as NetSuite, QuickBooks, Sage, or SAP into DualEntry within 24 hours, against an industry norm where ERP migrations can take months and cost millions in consulting fees. The company also reports more than 13,000 third-party connections across banking, CRM, and HR systems, according to its public materials.

That foundation matters for DualEntry Tasks because agentic finance depends on access. An agent that can chase an invoice or code a transaction needs live connections to bank feeds, expense tools, and approval chains, not a chat window bolted onto a closed database. The company's October 6 two-way integration with Expensify, which syncs expense and corporate card data directly into DualEntry without CSV exports, shows the same pattern: deepen the data layer, then let agents act on it.

The broader message is a reframing of the product category. Where the company once marketed itself as an AI-native ERP, the DualEntry Tasks announcement describes the destination as an agentic ERP, positioning the software as an active participant in finance operations rather than a passive ledger, while keeping accountants in control.

Finance is becoming the new agentic battleground

DualEntry Tasks is entering a crowded and well-funded fight. Microsoft is deploying agentic AI across Dynamics 365, integrating autonomous agents into Teams, Outlook, and Copilot to handle sales and service workflows, according to Futurum's reporting summarized this month by the AI Agents Directory. The same month, rival AI-native ERP vendors have been raising at pace: a 2026 comparison of AI-native ERP vendors published by ERP Research notes that Rillet raised a $70 million Series B co-led by Andreessen Horowitz and ICONIQ, while Campfire closed a $65 million Series B co-led by Accel and Ribbit.

The common thread is that agents are moving from chatbots that answer questions into systems that execute multi-step work inside the tools companies already run. Similar dynamics are playing out across enterprise IT, where governed agent frameworks are being layered over mission-critical operations, a trend covered recently on this site in the context of DXC's partnership with LangChain to scale governed AI agents, and in scientific research, where autonomous agents are being built to run experiments.

The open question for DualEntry Tasks is the same one facing every agentic finance product: trust. Finance teams hand agents access to money movement, journal entries, and audit-sensitive records, which is why the product's emphasis on audit trails, approvals, and human-in-the-loop control is as important as the automation itself. Agents that silently post entries will not survive a real audit; agents that show their work might.

For now, DualEntry Tasks is positioned as the workspace where that trust gets exercised daily: a task board where agents propose and complete work in the open, and accountants approve or redirect. If the agentic ERP vision behind DualEntry Tasks holds, the month-end close could stop being a monthly event entirely, replaced by a finance operation that simply never stops moving.