How former KKR CIO Ed Brandman is using generative AI to turn messy data rooms into real-time intelligence—and why top firms are already on board.
by Dan Costa | Republished from Worth. View original article.
Private-equity diligence isn’t just a reading assignment; it’s a race against the clock across thousands of pages that rarely look the same twice. Ed Brandman, the former KKR partner and CIO, believes that’s precisely the kind of messy, high-stakes terrain where specialized AI is best suited. His company, ToltIQ, ingests everything you’d typically park in a virtual data room—purchase agreements, quality-of-earnings reports, customer cohorts, side letters—and turns the pile into queryable intelligence for deal teams under time pressure. The pitch is blunt: let machines do the sifting so humans can do the judging.
Brandman’s framing of the problem is informed by scar tissue. He spent more than a decade developing technology and data strategy at KKR; before that, he held senior roles at PwC and Robertson Stephens, co-founded a trading-tech company, and helped launch early FIX connectivity at J.P. Morgan. In short, he’s shipped systems inside the institutions that now need better tooling.
“We put together a team with deep PE and credit industry knowledge and amazing technical engineering skills,” he says. “Our goal is to change the way many diligence activities are done at GPs and LPs.”
If the name sounds unusual, that’s intentional. The “tölt” is a smooth, efficient fifth gait of Icelandic horses—a metaphor Brandman borrowed from a performance review at KKR that stuck with him.
“The name ToltIQ was inspired by a moment during my time at KKR… Icelandic horses have an additional gait, the tölt. It’s smooth, fast, and efficient… My boss was asking me to find that ‘extra gait,’ that fifth gear,” he says. “ToltIQ embodies this by helping investment teams move at extraordinary speed with clarity and grace.”
What does that look like in practice? The platform is vertically focused on post-signing, VDR-centric diligence. Teams upload document sets; ToltIQ analyzes, categorizes, and links them; then analysts, associates, and partners can interrogate the corpus with targeted questions—“Where are revenue recognition risks in the top five contracts?” or “Reconcile cash conversion vs. peers over 24 months”—rather than spelunking through nested folders. Benchmarks achieved productivity gains of 35% to 85%, with some tasks reducing from hours to minutes and multi-week projects compressing into days.
Those are big claims, but they align with what Worth is hearing across the diligence advisory ecosystem: the lift comes not from a single summary, but from many small accelerations in triage, cross-references, and exception finding.
The model strategy is pragmatic rather than ideological. ToltIQ runs on “frontier models from OpenAI and Anthropic,” and the team publishes regular evaluations comparing model performance on finance-specific tasks—from reasoning to precision and output density. The point isn’t to anoint a champion so much as to route each query to the model most likely to handle it well, and to keep switching as the state of the art advances.
Brandman’s line on this is understated but telling: “Generative AI success at this stage is a combination of art, science, creative destruction of your codebase, endless research, and measured risk-taking.” In other words: expect churn under the hood; the UI should remain intuitive (not always obvious in the streaming world of generative AI.)
Momentum matters, and the financing suggests real demand. Earlier this year, the company raised $12 million in a two-tranche Series A led by FINTOP Capital and JAM FINTOP. In announcing the round, Brandman called the product a “co-pilot for investment teams,” emphasizing that the aim isn’t to replace associates but to give them leverage on the clock. ToltIQ is already in use across dozens of GPs, LPs, family offices, and diligence firms—including notable names like HarbourVest Partners, Fortress Investment Group, Investcorp, and PPC Enterprises.
Rebrands are often cosmetic, but the company’s name change from DiligentIQ to ToltIQ signals tighter product positioning. The emphasis is on workflow over wizardry: secure ingestion, governance, and permissions; repeatable prompts tailored to private-market documents; and line-of-sight from findings to footnotes. For deal professionals, that last piece is the differentiator. Nobody wants an LLM’s confident summary that can’t be traced back to the PDF it paraphrased. The platform’s thesis is that trust stems from being able to click on a sentence and land directly on the source.
There are reasonable caveats. Every firm’s document hygiene is different; messy inputs still require human adjudication. And the model landscape is volatile: what beats a baseline today may be average next quarter. Brandman appears comfortable with that reality. He’s explicit about staying “nimble” as models evolve, and his team’s steady cadence of comparative analyses suggests they know the only winning move is continuous re-evaluation. That mindset—not a particular model—may be the real moat.
What makes ToltIQ worth watching isn’t that it’s an AI company; it’s that its design assumptions are native to private markets. The product doesn’t ask a buyout team to change how it works so much as to shift where the effort goes—from hunting for needles to arguing about what the needles mean. If Brandman is right, the “fifth gait” isn’t speed for its own sake. It’s a smoother motion across rough ground, allowing decisions to be made with more context and, crucially, more time. For investors, time is alpha. For everyone else in the deal, it’s sanity.
© 2025 ToltIQ
