How Benchmark is Building the First-Party Data Engine for $1 Trillion in Assets

Benchmark is an AI-native investment platform used by firms managing ~$1T in assets under management, including some of the world’s top financial institutions. They help teams manage the entire deal flow while continuously building institutional knowledge—turning every document and decision into a long-term competitive edge.

Benchmark originally launched out of Y-Combinator as BidSight with a platform to help real estate developers manage projects and data. But the real opportunity revealed itself in investing, where the same document chaos—and much higher stakes—existed at every stage of diligence.

They pivoted quickly, applying their backgrounds in finance to help professional investors screen deals and manage unstructured files. They quickly realized, most deals don’t close. But even those contain valuable insights that usually disappear—locked in analysts’ heads or lost in disconnected folders. Benchmark saw the opportunity to change that: capture and normalize first-party data from every deal, closed or not.

But as they leaned into this idea, a new friction emerged.

“Document parsing was a bottleneck at day zero. Once we switched to working across the full spectrum of investment documents - presentations, excels, contracts, scanned PDF's, etc - it became obvious we wanted to partner with experts to solve that layer.” - Connor Jansen, Co-founder of Benchmark

They tested open-source tools and built their own, but nothing held up even after spending precious engineering resources on it. After choosing Reducto, they never looked back—now on track to process over 3.5M pages annually with ease.

Real-world parsing, not just a demo

Benchmark evaluated a wide range of vendors. Some looked good on the surface with promising benchmarks, but most broke down under real-world conditions: messy Excel spreadsheets that were thousands of rows long and inconsistently laid out.

“Excel handling was the big differentiator. Other providers would show strong benchmark scores, but when we ran holdouts with real-world messy files, they’d fall apart. The demo they would show would be on a really well formatted table on a single Excel sheet.”

What set Reducto apart wasn’t just accuracy—it was the ability to scale without friction. The API delivered clean, structured output with traceable citations, no matter the volume. Whether it was a handful of files or tens of thousands of pages, the performance held steady and brought peace of mind to Benchmark’s small team.

“It’s been super nice to not have to worry about a customer onboarding spiking our document processing volume unexpectedly – I just don’t need to worry about that with Reducto.”

The Document Builder: Completing Hours of Work in Minutes

One of Benchmark’s breakout features is its Document builder—a Word-like interface for generating and editing any form of documents like investment memos, research reports, portfolio updates, and more.

The Document Builder lets users choose which files the document can reference and upload or select previous documents to inherit their style and tone. Benchmark then creates an editable outline and generates the actual document. A user at a large multi-strategy fund noted it reduced their IC material creation process from a week to less than 2 hours.

Using AI to generate documents only works in heavily regulated industries like finance if every claim is backed by a source. Benchmark’s documents are built with that in mind—each claim is linked to its original citation, so users can verify exactly where the data came from before making a decision.

"One of our key design principles is always showing citations directly alongside all of our outputs. We can do this well because Reducto provides best-in-class chunks, which often require no post-processing."

Scaling Document Workflows Without Scaling Headcount

Today, Benchmark is on track to process over 3.5M pages a year through Reducto’s infrastructure—automatically processing, chunking, and routing files from user uploads and forwarded emails with virtually no manual overhead. For a small team of 6, that’s the difference between treading water and moving fast.

"Before we started Benchmark, we spent weeks trying to handle heavy obstructions like architect signatures and watermarks on documents. If we'd had Reducto back then, it would have been a one-day solve"

With parsing fully offloaded, the Benchmark team can stay focused on building what matters—building and deploying AI agents that solve real problems across the entire investing lifecycle. Reducto doesn’t just handle the plumbing; it gives the team back their time.

And when something does break—or a novel edge case comes up—they don’t get stuck debugging someone else’s API. They reach out and hear back fast.

“Having founder-level access and also the response times to our crazy edge cases is super meaningful because it unblocks our work. They treat everyone like they’re the most important customer in the world.”

For Benchmark, Reducto isn’t just infrastructure. It’s what makes their small team feel ten times larger.

What’s next: A Rich First Party Data Source

​​Benchmark isn't just speeding up workflows—they're building a long-term advantage. By parsing every document through Reducto, Benchmark's AI agents can understand any document—enabling them to quickly construct structured datasets and analyses across every deal a firm evaluates, generating "a really, really rich first-party data set specific to that firm, allowing them to constantly learn, improve, and compete with no data loss deal to deal," as co-founder Connor Janson puts it.

That foundation is enabling the team to push into more ambitious territory—using the rich first-party datasets Benchmark constructs to answer truly complex queries like “What types of deals do we consistently take longer to decide on, and is that decision speed correlated with outcomes”.

And as models continue to improve, the roadmap only expands.

“We're really excited about the latest model generations, especially longer context models—they unlock a lot of product surface area that wasn't previously achievable at our production standards.”

Behind that velocity is a key advantage: the Benchmark team has lived the workflows they're building for. With backgrounds in private equity, investment banking, and data-room tooling, they're able to move from insight to production quickly—because they already understand what the end user needs.

As the infrastructure powering those insights, Reducto helps Benchmark scale to unlock a future where every firm can move faster and get smarter, deal over deal.