Ian Crosby is back. The founder whose bookkeeping startup Bench Accounting collapsed in 2024 has just raised $10 million in seed funding for a new venture called Synthetic, according to TechCrunch AI. Khosla Ventures led the round, with Basis Set Ventures and Shopify CEO Tobias Lütke joining in.
The pitch is bold: a fully autonomous AI bookkeeper that produces accrual-based financials with zero human involvement. No accountants in the loop. No hybrid model. The product is still in the design phase, and Crosby openly admits the vision may exceed what today’s foundational models can deliver.
What stands out here is the willingness to bet on a founder most VCs would avoid.
Why Khosla took the swing
Khosla partner Jon Chu told TechCrunch AI he likes deals other investors flinch at. “I tend to run towards controversy a little bit,” Chu said. “In controversy, groupthink often shapes the narrative rather than the truth of the story itself.”
He pointed to Parker Conrad as precedent. After getting pushed out of Zenefits in 2016, Conrad went on to build Rippling, now valued near $17 billion. The message: a public flameout doesn’t disqualify a founder.
Crosby’s own version of the Bench story matters here. He says he was fired by the board in 2021, three months after turning down a $250 million acquisition offer from Brex. The board disagreed with his strategic direction as cash burned. Bench’s new management couldn’t right the ship, and the company imploded years later.
“He took a big swing, made a few mistakes. That didn’t go well,” Chu said. After leaving Bench, Crosby did stints at Shopify and founded Teal, which Mercury acquired 18 months in. Chu says he checked references from that period and “they all had fantastic things to say about Ian.”
The product gamble
Synthetic is going all-in on full automation. Most accounting startups, like Xero, still lean on human accountants behind the scenes. Crosby refuses to ship anything halfway.
“We’re not going to release anything that’s not fully autonomous,” he told TechCrunch AI. “It’s that or bust.”
The initial customer base is narrow on purpose: AI startups and other software companies. Even so, Crosby concedes today’s models still make meaningful bookkeeping errors. He used a self-driving car analogy to describe the scaling risk:
- The prototype works on one “street” (a narrow customer profile)
- Whether it generalizes to any “street” (broader customers) is unknown
- “We haven’t driven down enough streets to know if it’s going to crash”
His answer to that uncertainty? Money and time. “I’ve raised years of cash, so we can just wait it out,” Crosby said. The plan is to sit tight until foundational models get reliable enough for accrual accounting math.
Why this matters
Three things make Synthetic worth watching beyond the founder drama.
First, bookkeeping is one of those workflows everyone assumes AI will eat, but nobody has cracked it end-to-end. The accuracy bar is brutal. One wrong journal entry can compound across an entire fiscal year.
Second, Crosby is making a directional bet that most AI services companies are dodging. The hybrid AI-plus-humans model is winning revenue right now. Full autonomy is a longer, harder play that depends entirely on model capability curves continuing upward.
Third, Khosla’s reasoning here is a tell about how top-tier VCs are thinking about founder risk in 2026. Track record matters, but so does the post-failure arc. Crosby’s three roles after Bench gave Chu enough signal to write a check most firms wouldn’t.
For practitioners building AI services businesses, the question Synthetic is implicitly asking is the one worth sitting with: do you build for the model you have today, or the model you think you’ll have in two years?
Full details at the original TechCrunch AI report.