Rogo hits 27x ARR growth using OpenAI to save bankers 10+ hours a week

The grind of junior banking analysts — the endless nights combing through SEC filings, pitch decks, and private data rooms — has been a rite of passage for decades. Rogo, an enterprise AI platform that just finished its first year out of stealth, is proving that ritual may finally be obsolete. The company reports it now serves over 5,000 bankers across public investment banks and mega-cap private equity firms, saving analysts more than 10 hours per week on meeting prep, company profiling, and market research.

Tumas Rackaitis, Rogo’s co-founder and CTO, doesn’t mince words about the stagnation he saw. “There hasn’t been a company upending finance workflows for decades,” he says. “We built Rogo to be the essential tool every investment banker relies on daily — just like Bloomberg but for deep financial insights.” That ambition is backed by a 27x jump in annual recurring revenue, powered by a layered architecture of OpenAI models fine-tuned on financial datasets from S&P Global, Crunchbase, and FactSet. The platform now searches and analyzes over 50 million documents.

The model strategy is pragmatic rather than flashy. GPT-4o handles chat-based Q&A and heavy financial analysis. o1-mini structures data for search, and the full o1 model is reserved for evaluations, synthetic data generation, and advanced reasoning — the high-stakes stuff hedge funds and private equity firms actually care about. A team of former bankers and investors labels datasets to keep outputs grounded in reality. As Rackaitis explains, the most complex insights get distilled into smaller, faster models for high-velocity environments like investment banks.

Rogo’s recent hire of Joseph Kim as Head of AI signals where this is heading. Kim joined from Google’s Gemini team, where he focused on reinforcement learning from human and machine feedback. That expertise suggests Rogo is serious about tightening the feedback loop between model outputs and the financial professionals who rely on them daily. The question isn’t whether AI can parse a 10-K — it’s whether the industry’s old guard is ready to trust it with the answer.