The New York firm cut a preliminary investment overview from two hours of analyst time to about 20 minutes, with every figure footnoted back to the document it came from.
V7 Go is the platform 10 East runs the top of its investment funnel on. The New York firm gives its members access to private market opportunities across private equity, private credit, real estate and niche strategies, and it looks at hundreds of deals a year to find them.
A preliminary investment overview used to take an analyst one to two hours. It now takes about 20 minutes, a 60 to 70% reduction, and every figure in it opens the document it came from.
Hundreds of deals a year, 10 to 15 every week
10 East has grown quickly over the last four to five years, and the investment team’s answer to that growth was to look at which parts of its process were repetitive rather than analytical.
The first pass review was the obvious candidate. Every incoming deal needs a short, high level overview before the team can discuss it, and at 10 to 15 new deals a week, assembling those overviews by hand was consuming a serious share of the team’s week.
“We use V7’s AI tools to do a very quick review and create a first pass so we can have those discussions effectively with standardized materials,” says Louis Mayer, Managing Director at 10 East. “Those always have human review and are augmented by our team, but we can automate the process of templating and creating some of those materials.”
Two hours to 20 minutes
Mayer is precise about where the saving comes from, and about what it is not.
“Every new investment overview that we put together probably takes anywhere from an hour to two hours to review the information, create the template, put it together, refine it, review it, and have it ready for an overview conversation with the broader team,” he says. “We think the V7 tool reduces the time required from a human to pull one of those preliminary reviews together down to about 20 minutes per overview, and that’s a 60 to 70% reduction in time spent on each one of those overviews.”
The hours do not disappear from the firm. They move. “So we can cover a lot more ground with a lot less time and harvest those resources for better use,” Mayer says.
Further down the funnel, once a deal is approved for full diligence, 10 East produces a 25 to 35 page investment memo. The deep dive underwriting stays with the team. The repetitive parts, such as bios in a templated style and firm overviews pulled from input materials or public sources, do not.
Quarterly reporting on 300 portfolio positions
The same pattern holds after the investment closes. 10 East reports quarterly on 300 active line items, with similar style inputs and similar style outputs every quarter, and it is the kind of work that resisted conventional automation.
“There is no really good way to automate that historically because it required more human style judgment, and so it couldn’t be directly templated,” Mayer says. But the task is a good fit for a defined workflow that can transpose information from an input source into an output template in the right style, with the team reviewing the result.
Six days from first look to investment decision
The clearest single test of the change was a transaction with almost no runway.
“There was a transaction that we did where we had probably six days to go from finding out about the transaction to pulling together information to starting our process to make a decision and collect commitments,” Mayer says.
“That would quite literally not have been possible before we had access to these AI tools, because it would have taken us far too long to adjust the materials that we received and create the Investment Committee materials we needed in order to approve a transaction and move forward to make an investment.”
Why auditability decided which tool survived
10 East is SEC registered, so compliance and confidentiality were the first filter: the firm needed assurance that the underlying models are not training on its data and that its information stays secure.
The second filter was auditability, and it is the one Mayer returns to.
“If you don’t know where the information that’s being exported from the AI tool comes from, you have no idea if there’s any accuracy to it,” he says. He has heard the stories: outputs asked to track KPIs, where the numbers were simply invented. “If you can’t source the data back to their root source data, you can’t believe any of it.”
Every value in V7 Go traces back to the material it was drawn from through visual grounding. 10 East took it a step further and worked with V7 to build custom footnotes into its own output documents.
“In the output document, there’s a tag back to the underlying source document, both in a static form but also live that we can click directly to see exactly where it came from,” Mayer says. “That creates an audit trail that we would not be able to function without.”
An NDA review workflow, built by someone who does not code
The investment team were the early adopters at 10 East. The firm is now extending the workflows to legal, compliance, finance, operations and client service.
One of those was built without any help from engineering. Mayer built an NDA review workflow himself: 10 East’s standard template for what it is willing to accept and reject, with the output producing a proposed redline.
“We were quite pleased with ourselves, because using the V7 system I was actually able to build that myself with no coding skills whatsoever,” he says.
Treat the model like an analyst who needs training
Mayer’s advice to other firms is mostly about scope. The firms that fail, in his experience, start too big.
“They expect AI to do everything for them, to do it perfectly, and when it doesn’t, they get frustrated and they just walk away,” he says.
His alternative framing is a hiring one. “Think about the LLM as a bad analyst that needs to be trained. And once they’re trained, they get pretty good, but you’re starting with some relatively raw materials.” Guardrails, instructions, and time spent developing each workflow as if it were a new employee. Plus an internal champion who owns the relationship, and first use cases that are approachable rather than complex.
A framework for choosing what to automate next
10 East recently put a structure around that choice. The team listed every task that takes a lot of time, then ranked each one on three axes: how much it affects the bottom line, how clean and ready the source data is, and how ready the users are to adopt a tool for it.
“We ranked every single one of those use cases to determine which would be the most likely to impact our bottom line, actually be successful in development and implementation, and then actually be used,” Mayer says. Roughly 50 candidates went in. Ten came out as the list to build next.
Compressed to three questions, which is how Mayer puts it to other firms: what do you want to accomplish, is the data you need available, and will the users actually use the tool?
About 10 East
10 East is a New York based investment firm giving members access to private market opportunities across private equity, private credit, real estate and niche strategies, with more than $1.8 billion of invested capital. Founded by Michael Leffell.
Read the full story in 10 East’s customer story, or see how V7 Go is used across private markets and finance teams.
