Since joining Glilot, the question I hear most often has nothing to do with AI itself – it has to do with time. LPs ask whether exits will happen sooner. GPs debate whether holding periods will compress. Founders wonder whether AI lets companies scale and get acquired years earlier than before.

It’s a fair question. Diligence that once took months can now take weeks, and capital moves faster than it used to. But I’d frame the real question differently: has the speed of investing changed our expectations more than it’s changed the underlying work of building great companies? Sitting between LPs who fund our conviction and GPs who act on it, my answer, after many of these conversations, is yes.

What AI Has Actually Changed

AI has genuinely upgraded venture’s operating model. Founders build with remarkable efficiency; small teams now do what once required entire engineering organizations. Investors synthesize market research in minutes and evaluate competitive landscapes at a scale that would have been impossible a few years ago. Fundraising has shifted too – companies often arrive at first meetings with more polished products and clearer data, though this varies by sector and stage. On the LP side, reporting is faster, and questions that once took a week of pulling data can often get answered in an afternoon.

That’s really good. But it carries a risk. In our enthusiasm for faster ways of working, we can start assuming the outcomes themselves should arrive faster too. AI has compressed many of the activities surrounding venture investing without rewriting the mechanics of building businesses that endure. Confusing operational speed with value creation is one of the more subtle risks facing our industry right now.

What It Hasn’t
Venture has always run on asymmetry – a handful of exceptional investments can define a fund, and finding them has always required technical insight, conviction, and patience. AI strengthens those capabilities, but the qualities that separate extraordinary companies from merely good ones are still resistant to automation.

Enduring companies are built through thousands of decisions about product, hiring, customers, pricing, and execution that compound quietly, long before the market notices. AI changes how quickly founders reach each decision. It doesn’t change how long it takes good decisions to add up to an obviously great company.

Cybersecurity makes this vivid. It’s one of the fastest-moving sectors in tech, yet the process by which great cybersecurity companies become trusted partners to enterprises hasn’t sped up nearly as much. Large organizations still evaluate vendors carefully, and mission-critical infrastructure still depends on confidence that takes time to earn.

The same pattern holds more broadly: many companies now reach product-market fit faster than a decade ago, but becoming genuinely indispensable to customers tends to stay a long game. The companies that define categories rarely do so because they moved fastest in year one – more often it’s because they kept making better decisions than competitors, year after year.

Friction vs. Judgment

This is where the conversation inside venture is starting to shift. We increasingly celebrate speed as if it were the objective itself: how quickly diligence finishes, how fast capital deploys, how soon companies scale. These are useful questions, but not the most important ones.

Venture has generally rewarded firms not for the fastest decisions, but for the right ones. Faster diligence is valuable because it can lead to a better decision, not because it finishes sooner. AI-assisted sourcing matters when it surfaces founders who’d otherwise be missed, not just more of them.

What This Means for Founders, GPs, and LPs

For founders, there’s never been a better time to build – but the same tools are available to everyone, so speed alone is becoming less of a differentiator. The edge is shifting to what AI can’t replicate: customer understanding, technical originality, resilience, and the ability to earn trust over time.

For GPs, the logic is similar internally. AI should free up time for the work that actually compounds returns: understanding founders, evaluating technical differentiation, and helping portfolio companies through inflection points. The best investors have never been distinguished by how fast they gather information, but by how well they interpret it.

For LPs, this argues for a different set of diligence questions. Rather than asking how much AI has sped up a manager’s process, it’s more useful to ask how AI has improved the quality of their decisions, and whether the firm has a repeatable way of identifying exceptional founders. Speed alone can reward the wrong behavior – it’s possible to move fast and still make worse decisions. In my own conversations with LPs, the ones I particularly value don’t ask only about our impressive speed; they also ask about our conviction, and whether we can defend it years later.

The Discipline of “Not Yet”

The best firms should use every tool that makes them genuinely faster: admin, research, internal process, portfolio support. But that efficiency shouldn’t reduce the rigor applied to investment decisions, or rush an exit before a business has reached its potential. Some of the most successful venture-backed companies got there not by pursuing the earliest possible liquidity, but by continuing to build and pivot long after they had the option to stop.

The View from the IR Desk

Working in investor relations puts me between institutional investors and venture managers – one side focused on distributions, the other on building extraordinary companies. Both are right. The job isn’t choosing one perspective over the other; it’s resisting the temptation to assume that because everything moves faster, value creation does too.

My conversations with LPs are rarely just about performance. More often they’re about conviction and repeatability, and about how technology is changing not just the companies we invest in, but how we invest. I’ve come to think that’s what IR is really for – translating the realities of company-building to investors, and bringing the priorities of institutional capital back into the venture ecosystem.

AI will keep reshaping venture capital in ways we can’t fully predict. Some of those changes will make us faster; the best will make us smarter. The firms that define the next decade will likely be the ones that understand the difference – using technology to remove friction, not judgment, while holding onto what has always produced exceptional outcomes: intellectual honesty, disciplined decision-making, and real partnership with LPs and founders alike.

AI is driving an extraordinary rise in opportunities right now. Our deal flow has never been richer. In the last quarter alone, we closed 4 new investments. We are very picky about new investments, so for us, that is a very high number. Given that, I thought I would share more clarity on what we look for in a new investment.

Since we founded Glilot Capital in 2011, one thing has become very clear: the founders are in the center of everything we do, it doesn’t matter how much work we put into helping portfolio companies, we are not the ones to determine if a company will succeed or not, it’s all about the founders. This is why the identity of the founders is the most important part of any investment. We can talk about markets, technologies and returns, but in the end, every successful company we’ve backed was built by people who combined vision, discipline and persistence. Our job as investors is to be great partners to those people, focused, committed and hands‑on, from the first check to the last major decision.

We invest in cybersecurity and enterprise software from seed through growth, but our filter always starts with the entrepreneurs. When I look at a new cyber deal, I do not run through a mechanical checklist. I ask a few simple questions: who are these founders, why are they building this, and do we see a path to build something truly meaningful.

What We Look For

When evaluating a cybersecurity deal, the first thing we look for is a founding team that can build a huge business fast. The best founders we’ve backed lead from the front, stay close to the details, and are ready to do the hard work themselves, not just delegate. They usually bring a deep understanding of the problem space, often from intelligence units, security roles or building products at leading vendors, and they know how to translate that experience into a clear product and business.

We care a lot about persistence. Markets change, funding cycles come and go, and nothing moves in a straight line. Founders who stick it out through challenges, keep their heads, and continue to think creatively are the ones who ultimately build lasting companies. When we see that combination of character, experience and clarity, it matters more than any single feature in the product.

Of course, we also look at the market and technology. We focus on problems that represent large categories, not just nice features, and we pay attention to timing, whether a problem is becoming urgent for customers, not just interesting. On the technology side, we prefer depth over wrappers: architectures, detection methods or data advantages that are difficult to copy and can support real scale. These are the same fundamentals we apply across every cybersecurity deal we evaluate.

But even in these dimensions, we look at them through the lens of the founders. We ask whether this particular team is the right one to build in this specific market with this kind of technology, and whether they can attract the people and customers they will need along the way.

Why We Sometimes Say No

Saying no is as important as saying yes. Over the years, we’ve learned that when something feels fundamentally misaligned, it is better for both sides to pass early. Usually, that misalignment shows up around the founders and the story they are building.

Sometimes we meet teams targeting crowded spaces without a clear, sharp reason for why their company should lead the category. Sometimes the vision and the market size do not match, or the team is incomplete for the kind of company they want to build. For example, strong technology without a true business leader, or the opposite. And sometimes the difficulty is simply that the founder cannot explain the problem and solution in a way that is clear and convincing.

These are not theoretical criteria; they come from many years of working with entrepreneurs through good and bad cycles. When we decide to say no, we try to be direct and transparent, because honest feedback can still be useful for the founder’s next step, whether with us or with another investor.

What Happens When We Say Yes

When we decide to invest, we do not think in portfolio terms; we think in terms of this specific company and this specific partnership. We built Glilot to be a focused, value‑added fund, not a factory that jumps from one trend to another. That means deep involvement: helping with strategy, opening doors through our network of CISOs and executives, supporting hiring and go‑to‑market, and staying close through every major inflection point.

One of the things I am most proud of today is seeing founders come back to work with us on their second or third company. For us, that is the clearest sign that the way we choose deals, and the way we show up after we invest is working. In the end, our investment criteria in cyber deals can be summarized very simply: we look for great founders with real problems to solve, and we commit to being the kind of partner those founders deserve.

If you’re building in cybersecurity or enterprise software and think Glilot Capital could be the right seed partner, we’d love to hear from you, reach out to our team.

For years, a proof of concept was exactly what its name suggested: an opportunity to prove the concept. If your technology solved the customer’s problem better than the alternatives, there was a good chance you would win the deal. The evaluation was largely technical, and technical superiority usually translated into commercial success.

Enterprise buying has changed.

Over the last few months, I’ve spoken with dozens of founders, CISOs and enterprise security leaders about why technically brilliant startups struggle to convert successful POCs into long-term customers. Some conversations focused on AI security, others on identity, browser security, endpoint protection or cloud infrastructure. The technologies were different, but the pattern was remarkably consistent.

That’s an important distinction. It changes how founders should think about enterprise sales, and how enterprise security teams should think about evaluating innovation.

Why Founders Think a POC Means the Deal Is Won

From the founder’s perspective, the buying journey feels logical:

1. Problem identified
2. Product built
3. Capital raised
4. Introductions secured
5. Security architect aligned
6. Platform engineering engaged
7. IAM lead focused on integrations

Eventually, you hear the sentence every founder wants to hear.

“Let’s run a POC.”

At that point, it feels as though the hardest part is behind you.

The assumption is simple: if technology performs, the deal should naturally follow.

For a long time, that wasn’t an unreasonable assumption. Enterprise cybersecurity rewarded technical superiority. If your solution delivered stronger detection, lower operational overhead, richer telemetry or integrated more cleanly into an existing security stack, there was a reasonable expectation that the better product would win.

Increasingly, that’s only half the story.

What the CISO Is Actually Evaluating

While the founder leaves the meeting thinking about technical differentiation, the CISO walks into another meeting thinking: “Great guys, good technology, but does my organization have the capacity to deal with that?”

The Seven Questions Behind Every POC Decision

The conversation inside the enterprise rarely begins with the product itself. Instead, it begins with a different set of questions:

1. Is this one of the most important problems we need to solve this year?
2. Are we already halfway through implementing something similar?
3. Will one of our strategic platform vendors add enough functionality over the next twelve months that introducing another vendor no longer makes sense?
4. Do we have platform engineers available to deploy another sensor, connector or integration?
5. Will our security architects have time to review another architecture, another data flow and another privileged access model?
6. If this POC succeeds, do we actually have the operational capacity to deploy it across the organization?
7. Do we have the processes, tooling and internal support to operate yet another solution?

Notice how few of those questions are really about technology.

They’re about everything surrounding the technology.

One security leader described a startup whose product genuinely impressed the evaluation team. The technology was stronger, the deployment model was cleaner and the roadmap was compelling. The POC still stalled.

Not because the technology failed, but because the organization had already committed itself to another strategic initiative. Engineers had spent months implementing it. Architects had signed off on it. Procurement was already well underway. Internal champions had invested political capital getting it approved. Even if the new product was objectively better, changing direction carried a cost the organization wasn’t prepared to absorb.

Another security leader described pausing an entirely different category of products. Again, technology wasn’t the issue. Engineering resources were already committed elsewhere, architecture teams were stretched, and there simply wasn’t a credible path from a successful POC to a successful production deployment.

You’re Not Competing Against Another Startup

Across these conversations, one theme kept emerging. Founders believed they were competing against another startup.

In reality, they were often competing against engineering bandwidth, implementation timelines, governance, procurement, projects already underway and, increasingly, the expectation that an incumbent platform would eventually deliver a “good enough” version of the capability.

What Founders Should Do Differently: Qualify Readiness, Not Just the Problem

The findings point to a different challenge than most sales methodologies acknowledge.

Founders don’t just need to qualify the technical problem. They need to qualify the organization’s readiness to change. And qualify it hard and early.

That means understanding what strategic initiatives are already underway, where engineering capacity is already being consumed, whether the customer is waiting for an incumbent platform to close the gap, and what would actually need to happen for a successful POC to become a production deployment.

These aren’t objections to overcome.

They’re realities to understand.

If your internal champion has already invested months backing another initiative, pretending it doesn’t exist won’t make it disappear. Help them build the business case for changing direction. Help them quantify the operational value. Help them explain why changing course creates more long-term value than continuing with yesterday’s decision.

Because if you’re asking an enterprise to replace an existing initiative, you’re not simply asking them to buy different software. You’re asking them to revisit architecture decisions, engineering effort, procurement work and months of organizational momentum.

The founders who consistently succeed aren’t just better at selling technology.

They’re better at helping organizations navigate change.

What CISOs Should Do Differently: Evaluate Without Overcommitting

The same conversations point to a challenge inside the enterprise.

Innovation is moving faster than enterprise evaluation processes were designed to handle. Security teams are now being asked to assess AI-native startups, new identity models, browser security, runtime protection and entirely new categories of technology, often using governance processes built for a market that moved much more slowly.

Maintaining high standards is essential, but every lengthy evaluation has an opportunity cost. Every six-month POC consumes engineering capacity, architecture reviews, operational attention and executive sponsorship. The longer those resources remain tied up, the harder it becomes to evaluate the next wave of innovation.

The challenge isn’t to lower the bar.

It’s to build an organization that can evaluate innovation rigorously without making every evaluation a major organizational commitment.

The Real Constraint: Change Capacity, Not Innovation

The conversations behind this all pointed to the same conclusion.

Enterprise buying hasn’t become harder because founders are building worse products or because CISOs have become more risk averse. It’s become harder because innovation has accelerated while an organization’s ability to absorb change hasn’t kept pace.

For founders, success now depends as much on understanding the organization as it does on understanding the technical problem. The best founders don’t just prove their product works. They help customers understand how to act on the outcome.

For CISOs, the challenge is different. As innovation continues to accelerate, the organizations that consistently identify the next generation of category-defining companies won’t necessarily have larger budgets or bigger security teams. They’ll be the organizations that become better at evaluating new technology without overwhelming the people responsible for deploying it.

That raises an interesting question.

If the ability to evaluate innovation is becoming a competitive advantage in itself, what should a modern enterprise evaluation process actually look like?

AI has become part of my daily work.

I use it to build presentations, analyze campaigns, challenge event concepts, sharpen messaging, research markets, and turn rough ideas into something tangible much faster. I regularly move between ChatGPT, Claude, Gemini, and Perplexity, not because one is necessarily better, but because each is useful for different tasks.

AI helps me do more in less time. It gives me the ability to explore more directions, move from idea to execution faster, and increase the output of a lean marketing team.

I also see its limitations every day.

AI confidently gets facts wrong, produces generic copy, misses important context, and occasionally destroys a perfectly good design. It can generate something polished in seconds, but polished does not necessarily mean interesting, accurate, or effective.

Recent Gartner research on how marketers successfully use generative AI reinforces what I have experienced firsthand: the teams creating the most value are not necessarily using the most tools. They are the ones changing how they work, while understanding where technology ends and human judgment must begin.

Here are five principles that I believe make the difference.

1. Start With the Problem, Not the Tool

“We need to use AI” is not a strategy.

The starting point should be a real problem: research takes too long, content is not being reused effectively, outreach cannot be personalized at scale, or valuable knowledge is scattered across the organization.

Most marketing teams began with relatively simple tasks such as drafting posts, translating content, suggesting subject lines, or summarizing documents. These are useful experiments, but they are only the beginning.

The real value comes when AI solves a recurring business problem, not when it simply gives the team another place to type a prompt.

2. Use AI to Think, Not Only to Produce

Some of the most valuable work I do with AI never becomes external content.

I use it to challenge assumptions, identify missing perspectives, compare different approaches, and argue against my preferred direction. When planning an event, for example, it can help me question whether the concept is genuinely distinctive, whether the audience mix makes sense, or whether the value proposition is strong enough.

For a lean marketing team working across events, content, community, portfolio support, partnerships, and brand, this creates significant leverage. We can test more ideas and move from a rough concept to real execution much faster.

But producing more is not the same as achieving more.

AI accelerates the work. It does not decide which work is worth doing.

3. Measure Outcomes, Not Output

Creating twice as many posts is not necessarily an achievement.

A presentation completed in half the time is only valuable if its message is clear. More personalized emails matter only if they lead to better engagement. More campaigns are not progress if they do not reach the right people or support the business.

Gartner’s research emphasizes the importance of establishing a baseline and measuring the actual effect of AI. That means looking at time saved, shorter review cycles, stronger performance, lower costs, and new capabilities, not simply the volume of content produced.

The most important question is not “How much did we create with AI?” It is “What became faster, better, or possible because of it?”

4. Keep the Human Touch

At Glilot Capital, our marketing is built around relationships and trust, with founders, investors, CISOs, portfolio companies, and partners across the technology ecosystem.

AI does not know the full history behind those relationships. It does not always understand why a message might work for founders but not for CISOs, the dynamics between partners, or when a technically correct answer is still the wrong thing to say.

That context still comes from people.

Anything published under our name needs a human owner responsible for its accuracy, quality, and tone. AI can support judgment, but it cannot take responsibility.

As content becomes easier to produce, the marketer’s value moves away from creating the first draft and toward asking the right questions, recognizing what is genuinely interesting, and protecting the credibility of the brand.

AI accelerates marketing. The human touch still makes the difference.

5. Create for People and for AI

One of the most important shifts highlighted by Gartner is that content increasingly serves two audiences: people and AI systems.

Customers may discover a company through Google, LinkedIn, an event, or a recommendation. But they may also ask an AI assistant to explain a market, compare vendors, identify experts, or recommend a solution.

This means content must be clear, structured, credible, and specific enough to be understood by both humans and machines. Traditional SEO alone is no longer enough. Brands need to become trusted sources that AI systems can recognize, interpret, and reference.

The fundamentals of strong marketing have not changed. Clarity, relevance, authority, and trust still matter. The way people discover them is changing rapidly.

Making AI Real

Soon, every marketing team will have access to powerful AI models. Access itself will not create a sustainable advantage.

The advantage will come from everything surrounding the model: proprietary knowledge, unique relationships, strong workflows, brand trust, and people who know when to trust the output and when to challenge it.

At Glilot Capital, Make It Real is how we think about turning ambitious technology into meaningful impact. The same principle applies to AI in marketing.

Making AI real is not about impressive demos, generic content, or adding another tool to the stack. It is about integrating AI into the way we think, decide, and operate, and connecting it to outcomes that genuinely matter.

AI gives us more speed, scale, and leverage. The human touch gives that power direction, context, and meaning.

That is where the real advantage is created.