Two years ago, a managing partner at a mid-market assurance firm told me that his team spent more time exporting journal entries to Excel than they spent actually reviewing them. The export. The pivot table. The VLOOKUP to the prior period. The manual review of whatever subset they had time for. Forty hours of audit work, maybe twelve of which touched anything that looked like judgment.
That conversation is why AuditPulsar exists. Today, we're announcing a Seed Round that will let us build the product that conversation demanded.
What the Funding Covers
Three areas, in order of immediate impact:
First, ERP integrations. We currently have production-grade connectors for SAP S/4HANA, Oracle NetSuite, and QuickBooks Enterprise. The next twelve months will add connectors for Microsoft Dynamics 365, Sage Intacct, and Workday Financial Management. These aren't superficial API wrappers — each connector handles the specific table structures, field naming conventions, and permission models of the underlying system. Building them correctly takes time we now have.
Second, expanding the ML training dataset. Our anomaly detection models are currently trained on 14.7 million historical journal entries drawn from engagements across manufacturing, retail, healthcare, and professional services. We're licensing anonymized ledger data from three additional data partners to push that number past 40 million entries by Q2 2026. Detection accuracy on held-out test sets is already at 97.3%. The incremental gains we're chasing are in edge cases — unusual account structures, non-standard chart-of-accounts configurations, multi-entity consolidations.
Third, hiring. We're adding four engineers, two of whom will work exclusively on the compliance and security infrastructure that enterprise accounts require. SOC 2 Type II re-certification is scheduled for March 2026.
What We've Learned in Production
The product shipped to its first paying customer in January 2025. We've had ten months of production use to observe how audit teams actually interact with the platform versus how we assumed they would.
The biggest surprise: teams don't use the anomaly scoring the way we designed it. We built the risk score (0–100) as a triage tool — start at the highest scores, work down. What we observed is that senior auditors use the score as a confirmation mechanism. They form a view of the high-risk population independently, then check whether the scores agree. When they don't, they want to understand why. That's a different workflow, and it's a better one. The next product update will add a "score reasoning" view that shows the three primary statistical factors driving any individual score.
The second observation: the PCAOB workpaper export is used on 100% of engagements where we expected it on maybe 60%. Firms are using it not just for PCAOB-registered audits but for any engagement where the client asks for documentation of the testing procedures performed. The format is clear, the evidence trail is complete, and it saves two to three hours of workpaper assembly per engagement. We'll add AICPA SSAE 18 format in Q1 2026.
The Problem We're Actually Solving
There's a version of this announcement that frames AuditPulsar as an "AI company" or an "automation platform." That framing is misleading. We're an audit tool. The AI is the mechanism, not the product.
The product is this: accounting firms have more journal entries to test than they have hours to test them. The professional standards require that testing to be documented. The documentation requirements create pressure to sample rather than test completely. Sampling creates coverage gaps. Coverage gaps are where misstatements hide.
AuditPulsar closes coverage gaps. The AI is how we do it at the scale and speed that makes complete-population testing practical. A firm using AuditPulsar doesn't tell a client they sampled 200 of 12,000 journal entries. They tell the client they reviewed the complete population and the platform flagged 47 items for human review. That's a different audit.
A Note on the Competitive Landscape
We're aware of the incumbent offerings from the Big-4 affiliated technology groups and several VC-backed startups. The relevant distinction is not feature parity — it's deployment model and data architecture.
AuditPulsar runs in your tenant. Client data never flows to our infrastructure. The anomaly detection models run inside the accounting firm's cloud environment, not ours. For most firms, that's not a preference — it's a requirement. Several potential clients have told us they would not use a product that processes client financial data on shared infrastructure, regardless of how strong the vendor's security certifications are. We built for that constraint from day one.
What Comes Next
The product roadmap for 2026 has two major milestones: the expanded ERP connectors mentioned above, and a workflow integration with the major audit documentation platforms — Thomson Reuters AdvanceFlow, CaseWare, and ProSystem fx Engagement. That integration will eliminate the export-import cycle between AuditPulsar and whatever workpaper system a firm already uses.
If you're an accounting firm considering a pilot, the fastest path is the Request Demo form on our site. We run a structured pilot program — 60 days, one engagement, your data, your workflow. We'll tell you what we find and what we think the product is and isn't ready to do in your environment.
Thank You
The firms that gave us access to their workflows, their data, and their candid feedback about what wasn't working — that's where the product comes from. We're grateful for that trust and intend to keep earning it.
The investors who participated in this round understand what it takes to sell software to accounting firms. They've been here before. Their involvement is operationally useful, not just financial.
We're not celebrating today. We're starting the next phase of work.
James McKinnon, CEO — AuditPulsar
Boston, MA — November 14, 2025