What’s 🔥 in Enterprise IT/VC #461
The one man $80M vibe coding platform post acquisition 45 days later 📈 🤯
Remember that one man AI startup that was sold for $80M to Wix?
Now it’s adding $1M every 2.5 days 🤯. Why? Insane product - removes complexity from LLM and power of distribution but NOT YET FROM WIX. 👇🏻🧵
More importantly one would think selling Base44 into the massive installed base of Wix would lead to $400K ARR added per day but it’s mostly product love ❤️ and viral.
What are some of his key takeways building and selling and continuing to grow Base44?
Guy Podjarny, co-founder of Tessl (boldstart is an investor) from the AI Native dev podcast summarizes a few here:
Here are a few more:
Base44's "Batteries Included" AI App Builder: Base44, an AI app builder, differentiates itself by providing a "batteries included" approach for creating full-stack functional applications, including user management, databases, email integrations, LLM leveraging, analytics, and logs. This enables both non-technical users (like restaurant managers or marketing professionals) and increasingly developers to build production-ready software, expanding software creation literacy to a wider audience.
Strategic AI-First Development: Maor Shlomo emphasizes that AI was critical for building Base44 as a single founder, allowing for exceptional velocity and outperforming heavily funded competitors. Key decisions, such as choosing Python for backend and JavaScript for frontend, were made because these are the languages LLMs are most proficient at, enabling easier and better code generation.
Overcoming LLM Limitations for Quality Software: To address LLMs' struggles with complexity and code quality (e.g., lack of refactoring), Base44 employs an opinionated architectural approach. This involves designing SDKs and services around what LLMs are good at, implementing guardrails, and using automated refactoring tests that prompt the LLM to refactor long code files, ensuring longevity and maintainability of applications.
Shifting Paradigms in Software Testing: Traditional unit tests were found to be a "headache" due to frequent code changes during Base44's early development. Instead, testing focused on high-level, end-to-end tests where an agent interacts with a browser, simulating a user to validate features. This approach, driven by AI agents, is more scalable and effective when implementation details change rapidly, providing a similar benefit to a team of manual QA engineers.
The Future: Broad Skills, Lean Teams, and Agentic Software: Looking ahead, Maor predicts developers will need to broaden their understanding across domains (frontend, backend, design, product) to effectively leverage AI agents for end-to-end execution, rather than specializing deeply in one area. He envisions a future of smaller, highly contextual teams. Furthermore, Base44 is investing heavily in building "agentic software," moving beyond current web applications to create systems with embedded AI agents that can research, automate, and enrich data, fundamentally changing how software functions.
The Moat: velocity and speed, taste of your product, distribution, brand and community - it will be copied but not just one attribute - need many to stay ahead.
Watch/listen to full podcast here:
Remember that other vibe coding startup Lovable? It’s getting inbound offers at double the last round week after closing at $1.9B (FT).
Lovable is really not affordable for anyone to buy save a few which means more seemingly expensive 1-5 person startups with some early traction could get Wix like offers which are absolutely game changing for founders when zero venture capital is raised. This one man acquisition is super important for incumbent SaaS vendors to understand. You don’t need an army of engineers to add significant AI value and don’t need to pay billions to move the needle 📈.
Here’s are a chart showing the dire straights many SaaS companies are in and why they need to get AI fast.
Before you blame the above chart on the Post ZIRP hangover, David chimes in:
* BTW... if you're wondering if this is just a 2021 GAAC hangover, it's definitely not. Prevailing median growth rates in public SaaS were over 30% every year since at least the early 2010s. It's also not just the effect of these companies getting gradually older/more mature. Virtually every public SaaS company, when looked at individually, saw its growth rate fall significantly in 2022-2025.If you want to be rewarded with an AI multiple of 10-15X forward NTM vs. 5-7x, then get on the AI train now. You can either build which is super tough for many of these slower moving SaaS cos or…buy the right one with the right founder and fit. Congrats to Wix and Maor as this is a huge lift for Wix. It goes to show that a one person team who has a vision with strong opinions can outrun and outbuild much better funded companies with the power of AI. Expect to see lots more of these smaller acquisitions to help dead in the water SaaS cos find new life.
And to answer Guy’s big question of how long will it take to build Base 44 with Base 55? The answer is shocking, 6 months, and if not, definitely less than a year. Think about that - insane complexity to build this which speaks to the steep improvement curve of AI and coding.
As always, 🙏🏼 for reading and please share with your friends and colleagues!
Scaling Startups
#🎯
#lessons from Elon as broken out by David Senra (Founders Podcast) 👇🏻 click for the rest!
Enterprise Tech
#market for devs bigger than ever
#and now here’s Grok Code Fast - users talking about speed
#on Grok Code fast
#pretty pumped for our portfolio cos Clay and CrewAI - both founding teams we partnered with at Inception or idea stage right at incorporation!
#🤯 when you look at it this way…now how to monetize all that spend - it’s coming!
#talk of the town this week…yes Google is cheap relatively speaking esp. if you see all that it is shipping below
#top 100 AI apps - “Google makes big moves For the first time, we ranked Google domains separately - and four properties made the web ranks: - @geminiapp (#2) @googleaistudio (#10) - @notebooklm (#13) - @googlelabs (#39)”
#folks finally waking up to the power of AI and agents - the more power you giveth the more it can killeth - A popular NPM package was hacked to run a prompt via Claude Code instead of normal malicious code. This trick avoids traditional malware scanners since it offloads logic to an LLM prompt rather than static code. The script scans local paths for secrets (e.g., wallet keys, env files) and exfiltrates them—all hidden inside a simple prompt. Detection tools relying on static analysis? Completely bypassed.
#if you’re interested in securing AI, then don’t forget to register for the inaugural AI Security Summit in October in SF - super 🔥 up to speak here
#no this won’t work…
#the pain and the opportunity - while AI tools promise speed, they often deliver bloated code, endless debugging loops, and cognitive overload. Tab completion remains the MVP—everything else? A mixed bag. Read 🧵 below 👇🏻
#Google crushing it in video - latest drop - image-to-video (Veo 3), AI avatars, and auto transcript trimming. Coming soon: noise cancel, filters, new export formats = full video pipeline inside Google. Startups in AI video creation & editing just got serious competition. Create something now at Vids.new
#and don’t forget to play with Nano Banana image editing as you can know be quite specific about what to change and keep - the use cases for advertising, marketing, fun are insane and more tools.
#just to give you an idea of what’s possible - 👇🏻 watch
#let’s not forget an overhaul of google finance with AI analysis built in…
#Stanford economists share new data that GenAI is quietly gutting entry-level job prospects for young Americans, especially in software and customer service roles. Using granular ADP payroll data, they found a nearly 20% drop in headcount among 22–25-year-old developers since late 2022—while older cohorts held steady. The twist? In fields where AI augments rather than replaces, young workers are actually gaining ground (WSJ)
#❤️ this Customer Success or CS Automation Engineers - here’s Nick Mehta one of early founders who started Gainsight, one of first and leaders in SaaS for the customer success category calling out portfolio co Clay’s GTM Engineering approach for success as well - - IMO, this is for most departments esp. in this AI-native world - engineering embedded in all functions
#Agents and robots to create over $920B of value - pretty insane estimate of value creation for the S&P 500 from Morgan Stanley 😲 highlighted in Barrons this past weekend - we’re talking $920B of net savings mostly from employee labor representing 28% of 2026 pre-tax earnings and 41% of Total Compensation Expense. This wil take a decade or more to play out…so still early!
#these case studies work! great to see port co Systalyze with this one…
#🤣 South Park Epiosde 3 Sickofancy…btw, this season is 🔥 - must watch!
Markets
#breaking down the Netskope IPO from Alex Clayton Meritech
#🤔 China does not want 4th best Nvidia chip
#more Israeli 🇮🇱 cybersecurity cos acquired - Okta buys Axiom for $100M + expands its cloud-based identity and permissions management capabilities (CTech)
Axiom was founded in 2021 by Itay Mesika and Ilan Dardik, who met during their military service in the Israeli air force in 2013. Since its inception, the company has raised $10 million, including a $7 million Seed round in December 2022, led by S Capital.
Axiom provides an identity-based approach to Permissions and Access Management (PAM), specifically designed for cloud, SaaS, databases, containers, and other virtual environments. Its platform enables organizations to implement Least Privilege principles, replace permanent permissions with Just-in-Time (JIT) access, and protect critical infrastructure without creating bottlenecks that slow business operations.