Tech1 hr ago

Ian Crosby’s Synthetic Raises $10M Seed for Agentic AI Bookkeeping

Synthetic, founded by Ian Crosby of Bench and Teal, secured a $10M seed led by Khosla Ventures to develop an agentic AI bookkeeping agent still in testing.

Alex Mercer/3 min/GB

Senior Tech Correspondent

TweetLinkedIn
Ian Crosby’s Synthetic Raises $10M Seed for Agentic AI Bookkeeping
Source: AndroguiderOriginal source

Ian Crosby’s new startup Synthetic secured a $10 million seed round led by Khosla Ventures to develop an agentic AI that handles bookkeeping. The system is still in testing and makes mistakes, but Crosby admits he’s unsure if the vision is achievable yet.

Context Ian Crosby, who previously founded the accounting platforms Bench and Teal, launched Synthetic in San Francisco to build an AI agent that can manage bookkeeping tasks. Agentic AI refers to software that can act autonomously to complete workflows, such as pulling data from banks, payroll, and invoices, then asking humans for clarification when needed. Crosby and a five‑person team have been working on the product since last year, aiming to replace manual data entry with continuous, real‑time reconciliation. His track record of building profitable SaaS businesses helped attract early‑stage capital despite the technical uncertainty.

Key Facts Khosla Ventures led the $10 million seed round, with angel investors including Shopify CEO Tobi Lütke and Basis Set Ventures. Crosby told reporters, "I don’t know if this is possible yet," highlighting the uncertainty around the technology. The AI bookkeeping system is currently in a crash‑test phase and still produces errors, which is notable because accounting requires near‑perfect accuracy for trust and compliance. The startup plans to integrate with major banking, payroll, and billing systems while keeping a human‑in‑the‑loop for ambiguous entries. Early tests show the agent sometimes misclassifies expenses or misses duplicate transactions, prompting the team to refine its entity‑extraction and reconciliation logic.

What It Means The funding shows that investors believe experienced founders can tackle high‑risk, technically demanding projects like agentic bookkeeping. Success will depend on proving the AI can reconcile transactions accurately, generate auditable trails, and handle exceptions without human intervention. Technical hurdles include robust connector reliability, deterministic classification of financial events, and limiting model hallucination that could misstate ledgers. Companies pursuing similar tools often combine machine‑learning components with rule‑based verification to satisfy auditors and reduce compliance costs. For now, the startup must improve reliability while demonstrating real‑world pilots and security controls. Demonstrating SOC‑2‑type controls and transparent error‑escalation processes will be critical to win confidence from accounting firms and enterprises. What to watch next: evidence of enterprise pilots, third‑party audits, released connectors to major banking and payroll platforms, and published accuracy or reconciliation metrics on live data.

TweetLinkedIn

More in this thread

Reader notes

Loading comments...