Ava is the first AI BDR to run outbound end to end, and you decide whether she runs autonomously or on copilot.

She finds leads or ingests accounts from your CRM, enriches them, sends personalized emails on behalf of your reps, follows up, handles replies and books meetings. Website visitor de-anonymization, intent signals, and a parallel dialer come built in.

A small team can manage her centrally for thousands of reps who never log in. Everything syncs two-way with Salesforce and HubSpot.

Ava runs outbound for companies like DoorDash and Grammarly, and one customer deploys her across 1,000+ reps. Ava is SOC 2 Type II audited, SSO and GDPR ready. Ava is how revenue teams grow pipeline without growing headcount.

Book a demo

​Big News!

We are now featuring 10 pitches and crowning first-, second-, and third-place winners. We wanted to give early-stage founders even more opportunities to pitch and present.

​The overall winner will still receive the highly coveted $39.99 Nano Pac-Man trophy in front of an audience of 150+ elite Bay Area founders and investors.

​All three get a feature in the VYS newsletter, which goes out to more than 7,500 founders, investors, and tech professionals at startups like Anthropic, Perplexity, Cursor, Scale AI, ElevenLabs, Writer, and Pinecone, plus Google, Amazon, and NVIDIA.

This section is popular. With so much happening in the SaaS, AI, and startup space, I will keep adding 4 or 5 links to what I’m reading or watching on weekly trends.

That’s right, we’re all GTM engineers now.

Teams are no longer limited to buying whichever software that’s the least terrible. They can build their own. A custom alternative used to require engineers, budget, security reviews, integrations, maintenance, and months of development. Now a team can build software around its exact workflow in days.

I stumbled upon this Retool survey of 817 people who created workplace software in late 2025. It found that 35% had already replaced at least one SaaS tool with custom software. Another 78% expected to build more custom tools in 2026.

Wait. Wut?

I know, crazy right? SaaS had a moat. It could make software faster and cheaper than customers could build it themselves. AI is eroding that advantage. Companies are no longer forced to pay for generic products filled with features they do not need simply because building something better is out of reach.

The software we are building might suck. But for the first time, building it may be easier than buying it. And what you make might suck less.

According to this survey, workflow automation is the most exposed category. Retool found that 35% of respondents had replaced workflow automation software. Internal administration tools followed at 33%, and business intelligence reached 29%.

CRMs and form builders came in at 25%. Project management reached 23%, and customer support reached 21%.

No category is safe.

I have even heard stories of teams that ripped out WordPress and vibe-coded their own website back end. When they want to optimize it for SEO and make changes, they ask Claude. It sounded crazy, but then I heard more stories about this. I haven’t tried it on my site, but want to.

When you look at the categories in this chart: Workflow automation, Internal administration tools, Business intelligence, or Forms, it’s not a random collection of SaaS products. These products are largely dashboards, permissions, and business logic sitting on top of a company’s own data.

That is exactly the kind of software AI can build today.

This chart measures where the technical barrier to replacement has already collapsed. When a company can generate the software itself, it’s not just the costs, but the lack of flexibility that I think is ultimately driving it. Plus, forced upgrades for features you never wanted and won’t use were always tough to swallow.

The most significant finding in the report is that software is being created without waiting for engineering or IT. Only 36% of Retool’s respondents identified as software engineers or developers. This would be unthinkable just a few years ago.

The rest worked across operations, product management, data, marketing, sales operations, IT, finance, business analysis, and other roles.

Retool found that 60% had built software outside IT oversight during the previous year. One quarter said they did it frequently.

Employees used to go around IT by signing up for unauthorized software. Now they can bypass IT by building it themselves. The most common reasons were, uh, hilariously obvious to anyone who worked in a big org. Building was faster. Existing products did not meet the need. The official IT process took too long. Shocking stuff.

These were not all junior employees experimenting with disposable prototypes. Retool reported that 64% of respondents were senior managers or above.

And of course, this creates real risks. A custom tool can expose sensitive data, break without warning, create maintenance work, or turn into another piece of software nobody fully understands. But those risks do not make the behavior disappear.

Companies will need to figure out how to govern custom software development because employees have already discovered and are building it.

I know you’re wondering what people are using to build all of this stuff with.

ChatGPT led the general-purpose LLM category at 70%. Gemini reached 56%, and Claude reached 53%.

The coding assistant category looked very different.

28% said they did not use an AI coding assistant. 72% used at least one coding assistant. GitHub Copilot led at 39%. Cursor followed at 35%. Codex reached only 12%.

That gap is fascinating. ChatGPT is the dominant general-purpose AI product among the people building workplace software. But OpenAI has not transferred that advantage to Codex.

Copilot and Cursor currently have more adoption inside the dedicated coding workflow. This matters because custom software isn't created through one magical prompt. People are assembling a stack. They might use ChatGPT to plan a tool, Claude to reason through a technical problem, Cursor to write the software, and another platform to connect it to company data.

The growth of custom software doesn't depend on one model or one application. An expanding collection of tools supports it by making every stage of software development easier.

So is SaaS going to disappear? Not all of it. Purchased software still has enormous advantages in reliability, security, support, maintenance, integrations, and compliance.

Keep in mind that we are in the experimental phase. Many custom tools and initiatives will fail. Others will become fragile internal systems that cost more to maintain than the subscriptions they replaced.

But SaaS companies can’t assume that building an alternative is difficult. The new comparison is not simply your product against another SaaS vendor. It is your product against what the customer can build.

This changes things.

A SaaS product must be easier to buy, safer to operate, faster to modify, and less expensive to maintain than a custom alternative. But generic functionality becomes less defensible when customers can recreate it around their own data and workflows.

The most durable SaaS products will own something hard to rebuild. Distribution, proprietary data, network effects, compliance infrastructure, deep integrations, or a system of record that the company cannot easily replace.

Everything else is becoming vulnerable.

Build vs buy used to be a decision reserved for companies with large engineering teams and enormous budgets. Now the person making that decision might be an operations leader with ChatGPT, Cursor, access to the company database, and a SaaS contract they no longer want to renew.

SaaS used to compete with other SaaS. Now it competes with whatever we can build.

Source: Retool, The build vs. buy shift

🎙️ 064 Gregory and Paul Show - Who Controls the Speed of the AI Race?

In E64, Gregory and Paul unpack the chaotic debate over whether AI will kill humanity and question the pseudoscience behind “P doom” predictions. They examine new data showing that five American companies control 71% of global AI compute, challenging claims that anyone can unleash an AI apocalypse. They also debate regulation, product liability, the responsibility of AI leaders, and why Meta could emerge as a powerful fast follower.

I'm a former creative director, 3x head of marketing, and founder of Vibe Your SaaS. After two decades in Silicon Valley, I moved to the PNW and created the VYS Group, the premier global network for startup founders, operators, and investors.

Have questions? Reply to this email. I write everyone back.

If you think this newsletter is rad (of course you do), help other cool people discover it.

This is your personal referral link: {{ rp_refer_url }}

Share it in your newsletter, blog, emails, wherever, to refer others and get credit towards unlocking cool stuff like this sweet, sweet VYS Guide to Founder-Led Sales, and if you refer 3 people, you can get access to the VYS Slack.

1 referral for this free guide

3 referrals for VYS Slack Connect Access

{{rp_personalized_text}}