How Sidekick Is Solving Real Problems for LPs and Managers

Questionnaires are getting longer. Strategies are more complex. And the volume of data being exchanged between investors and managers continues to grow. According to Dasseti COO Liron Mandelbaum, teams on both sides are feeling the strain, and looking for smarter ways to keep up.

“Our industry has a scary trend happening… questionnaires are getting longer and more sophisticated… and both sides are drowning.”

Dasseti’s AI assistant, Sidekick, is designed to relieve that pressure, by helping teams work faster without overhauling the way they already work.

“What really gets them interested in our AI solution is that this solution is built into their current workflows… it saves a massive amount of time on critical tasks.”

Helping both sides of the table

Sidekick is built for both LP and GP use cases. On the LP side, it helps with document extraction, manager evaluation, and year-on-year analysis. On the GP side, it speeds up responses and reduces the internal back-and-forth that often slows things down.

“Quite often, responding to investors is a bit of an art… You can end up getting writer’s block. Sidekick helps get out of that, by drawing on actual data to produce consistent, appropriate responses.”

Structure over surface

One of the key differentiators in Dasseti’s approach is the emphasis on structured data something many AI-first platforms overlook.

“Data-first eats AI-first for breakfast. If you're focusing on some slick new chat… you're forgetting: where is all that insight I’m getting going to?”

Instead of leaving information buried in chat threads, Sidekick builds and enriches centralised company and fund profiles. That’s where the real value compounds.

“If everything I’ve ever done with this manager is inside a chat… how do I compare that year over year? AI-first doesn't scale when you think of it that way.”

Avoiding the common AI traps

Liron’s view is clear: AI can be incredibly useful if it’s used the right way. That’s why Dasseti avoids asking models to answer large questionnaires in one go.

“Each question influences the others. It starts to overthink. The trick is to isolate what you’re asking. One question at a time, with only the relevant documents. It’s more work for the AI, but it’s the best approach.” “LLMs are designed to give you what they think you want, not what you need. That’s why we ask it to search, not to think.”

AI that fits the process

Dasseti’s approach is all about putting Sidekick inside the workflow, not creating a loop where the user is constantly interacting with AI to get the right result.

“You can almost think of the human in the loop as someone running on a treadmill. That’s not scalable for due diligence or investor relations.” “There’s a workflow. The AI automates what it can, and the human steps in when needed. That’s how you scale.”

A track record of security

Security isn’t an afterthought - it’s embedded in the architecture. Sidekick runs on Microsoft Azure and benefits from a zero-retention agreement that prevents any client data from being stored in third-party systems.

“We’re able to provide our clients with zero retention when information is sent outside our firewalls. It’s used to provide an answer, and then destroyed.”

A system that gets better with use

One of the concepts Liron returned to again and again was what he calls “the flywheel effect.”

“Every time you run through a cycle… you’re building a proprietary database. And as the wheel spins and your data gets richer, it gets easier to spin.”

That same logic applies to investor relations. The more you respond through Dasseti, the more accurate and efficient future responses become.

What’s next for Sidekick?

Liron doesn’t believe the biggest changes will come from smarter models. Instead, they’ll come from removing friction at every step.

“The benefit won’t come from the models getting better at science or research. It’ll come from automating more of the process, so clients can get to the answers faster.”

That means everything from auto-gathering documents, to reducing the time it takes to produce investment committee memos, to unlocking deeper analytics with less effort.

No hype. No vaporware.

“Sidekick is not a roadmap. It's live. It's been deployed for two years. And it meets clients where they are.”