Poetic raises $50M to automate complex enterprise workflows at 99%+ accuracy, backed by Founders Fund and Kleiner Perkins

Poetic, an AI startup coming out of stealth, has raised $50M led by Founders Fund, with Kleiner Perkins and Behind Genius Ventures (First Harmonic) participating.

What Poetic does

Founder and CEO Markie Wagner is targeting the class of enterprise workflows that software has largely left untouched: anti-money laundering investigations, underwriting, fraud review, insurance claims. The common thread is that the rules governing these processes have never been written down. They live in the heads of people who've done the work for decades, layered on top of whatever formal documentation exists. Wagner's argument is that the written documentation typically captures around 20% of what actually governs how a process runs.

Poetic's approach is to surface that missing 80% iteratively. The system takes whatever operating procedures a company has, generates a step-by-step AI operating procedure written in plain English, then runs it in front of domain experts. Their corrections — the threshold is actually $1M, not $10,000 — feed back into the system automatically, in a loop that Wagner compares to data labeling rather than traditional consulting or process mining.

“Poetic's an AI system that will learn and execute super complicated processes in some of the biggest companies in the world at over 99% accuracy... We work on things like anti money laundering and underwriting and fraud investigations where every single step matters... You raised $50,000,000.”

Why the architecture matters

Under the hood, Poetic converts those English operating procedures into code. When conditions are stable, the process runs deterministically as code. When something changes — a column name shifts, a save button moves — the AI steps in, reads the English-language intent, and repairs the code rather than breaking. Wagner frames this as a deliberate middle ground between pure code, which is brittle, and pure agents, which are improvisational and accumulate errors over time. The claim is that this hybrid gets to the accuracy thresholds these workflows require.

That accuracy bar is the core commercial argument. Wagner says an 80% eval score is considered strong in many AI contexts but is simply unusable for underwriting. A single CEO told him 98.5% accuracy would still generate hundreds or thousands of remediation hours. The company's stated target is 99%+.

Customer mix and go-to-market

Poetic is starting at the top of the market. Named customers include SoFi, Chime, and AIG. Wagner's rationale is that building for simpler workflows first tends to cap what the system can ever represent, and enterprises won't run separate platforms for easy and complex processes. Winning on the hardest workflows means everything easier follows naturally.

Deployment relies heavily on forward-deployed engineers, drawn from firms like Palantir, Retool, and Scale. Wagner says the profile has shifted over time — the people having the most impact are those who can think about how a business should reorganize around a new process, not just engineers who can write code.

The timing

Wagner started in AI research at Stanford, then at Waymo and Google, before dropping out to consult inside large legacy businesses. The consulting work crystallized the problem: vast amounts of institutional knowledge encoded in operating procedures that people follow manually, untouched by software because any code built to automate them would break the moment anything changed. He says he deliberately waited for the underlying models to improve before building, staying in contact with researchers until they told him the models were ready.

Enterprise sentiment, in his read, has moved from uncritical excitement to a harder question about ROI. CEOs who were enthusiastic a year ago are now asking how to get returns from AI investments, and the realization is settling in that throwing an agent at a process without first capturing how that process actually works produces little. Wagner calls the underlying challenge "the great migration" — getting the rules that govern core business processes into a form that AI can touch and improve.