The way founders pitch has changed more in the last two years than in the decade before that.
Not because investors suddenly want more slides or because decks got longer. But because AI has fundamentally shifted what software can do, and therefore what investors are actually trying to understand.
If you are building an AI-native B2B SaaS company today, the bar is no longer about whether you use AI. Everyone does. The real question is whether you are building something that replaces work, not just supports it.
The real question is whether you are building something that replaces work, not just supports it.
In this post, I want to break down how we think about pitches today at Vendep. From the very first cold email, to the demo, to the deck, and finally what can safely be left for diligence.
This is not a universal playbook, but it reflects how SaaS investors are increasingly evaluating AI-native companies.
Building the relationship before you need it
The most effective founders do not treat the first investor email as the start of the process. They treat it as the culmination of one. By the time they hit send, the investor already knows who they are, has been following the company for months, and is confirming a conviction rather than evaluating a cold story.
Aim to make first contact at least six months before you plan to raise. Not to pitch – just to exist in someone's peripheral vision. A quick update when something meaningful happens. The goal is that when you are ready, the decision is easy for them.
What the first email actually needs to do
When that moment does come, founders often overthink what the email needs to do.
You do not need a perfect narrative, every feature explained, or a full market analysis. You need to show that you are building something that feels inevitable.
In practice, that means three things.
First, a clear category description. What workflow are you owning? Not what feature you built, but what work inside a company you are taking responsibility for.
Second, why now. What changed in the market, in technology, or in regulation that makes this possible today, when it was not five years ago?
Third, the team. Not a CV dump, but a simple answer to why you are the group that can realistically dominate this niche first, and then grow it into something much larger.
At this stage, your goal is not to prove anything. Your goal is to make it obvious that this is not just another point solution. That you are not building a tool that sits next to human work, but a system that can actually do that work, continuously.
If your product can run a critical workflow 24/7, across geographies, languages, or customer segments, that matters. If it ties directly to revenue generation, cost reduction, or competitive advantage, that matters even more.
That is what gets a reply.
The demo is where claims meet reality
Once you get invited to a call, expectations shift fast.
Most VCs will want to see a demo early. Not because we love demos, but because AI has made decks less trustworthy. Everyone can claim intelligence but very few companies can show operational depth.
Here is the uncomfortable reality:
Most AI startups are, at their core, LLM wrappers. There is nothing inherently wrong with that. Models are good. They are interchangeable. Prompts help, but they are not defensible.
So when we watch a demo, we are trying to understand one thing: are you just wrapping an LLM, or are you building a workflow that would be genuinely painful to remove?
That difference usually shows up in a few places:
- Do you handle real customer data in a way that requires ongoing customization?
- Do you adapt to how each customer actually works, not just to generic inputs?
- Do you integrate deeply enough that replacing you would mean re-architecting internal processes?
If the answer is no, then your customers could replicate most of the value with their own prompts and an off-the-shelf model. That is not an AI-native company, it is a co-pilot.
In a strong demo, founders can clearly explain the problem, the current workaround, and why their solution creates such a large improvement that going back to humans would make no sense.
We also want to see focus. A few core features that matter deeply, not a long list of shallow capabilities. If you leave a demo and we cannot clearly articulate what problem you solve and for whom, the product (or the story) is not ready.
Product clarity comes before market storytelling
After the demo, the conversation usually shifts into more familiar territory: market size, competition, pricing, and ICP.
The fundamentals have not changed. We still want to understand who you sell to, how much they pay, and what they use instead today.
What has changed is how much weight we put on ICP clarity.
We are seeing companies grow very fast very early. Sometimes to millions in ARR within a year. That sounds impressive, but it often hides a problem.
If most of that revenue comes from broad SMB adoption, heavy discounts, or non-ideal customers, it tells us very little about whether your core market actually works.
What matters is not total ARR. It is ARR from the customer segment you plan to scale.
If only a small share of your revenue comes from that ICP, then your growth story is still unproven, no matter how fast the top line looks.
That said, speed used well is a different thing entirely.
The best AI-native companies use speed itself as part of the sales motion. Customers buy into momentum. When a team visibly improves the product week over week, prospects stop evaluating a static product and start buying into a trajectory. Short sales cycles follow naturally.
If your pipeline or ARR grew materially during the fundraising process itself, mention that. It is one of the strongest signals you can give.
The difference between the two shows up in retention.
When you do have meaningful revenue, gross retention is one of the most telling metrics we look at. It separates companies that capture easy AI budgets (experimental spend from curious buyers) from those embedded in how a company actually operates.
High retention means customers restructured a workflow around your product. Low retention, even with strong headline ARR, means you are still in the evaluation phase.
That distinction matters more now than it did in traditional SaaS. Selling the first contract has never been easier. Earning the renewal is where the work is.
The deck is about how the business works
When we ask for a deck, we are not looking for inspiration. We want to understand how the business actually works.
That means a clear business model. How you acquire customers. How you price. How your go-to-market motion evolves over time.
We care about traction, but we care even more about direction. What milestones you are working toward, how much runway you have, and whether those milestones are realistic given the team and the market.
This is also where references and early cohort data start to matter. Not because the numbers are perfect, but because they reveal how well you understand your own ICP.
Messy data is fine. Vague data is not.
On pricing: we are seeing a clear shift away from seat-based models, and it is worth addressing directly in the deck.
The most interesting AI-native companies price against the value they replace, not the number of users who log in. In practice, that means competing against a headcount decision rather than a software budget.
If your model still maps to seats in a context where you are automating work rather than supporting it, be prepared to explain why – or reconsider whether that model is holding your positioning back.
What can wait until diligence
Founders often try to front-load everything.
You do not need to unpack every technical detail or internal process in the pitch. That is what diligence is for.
If we get to that stage, we will anyway go deep into data pipelines, customer usage, security and economics.
Your job earlier in the process is to show that those answers exist, not to present all of them at once.
The shift founders need to internalize
The biggest change in pitching today is not about AI buzzwords or slide structure.
It is about responsibility.
AI-native companies are no longer evaluated as tools. They are evaluated as operators. Investors are asking whether your product can take ownership of real work inside a company, and whether that ownership compounds over time.
If your pitch reflects that shift, the conversation changes. If it does not, you will feel friction at every step.
The good news is that founders who get this right stand out very quickly. The bad news is that surface-level AI is no longer enough.
And that is probably a healthy thing for everyone involved.
.jpg)
.webp)