Hey — It’s Nico.

Welcome to another Failory edition. This issue takes 5 minutes to read.

If you only have one, here are the 5 most important things:

• ANS Commerce, a SaaS platform focused on D2C brands, shut down — learn why below.

• An opinionated guide on which AI Model to use in 2025.

• Nvidia debuts Groot N1, a model for humanoid robotics.

• Virtual desktop startup Nerdio raises $500M.

• Everyone is talking about “vibe coding” — learn more below.

Let’s get into it.

This Week In Startups

🔗 Resources

An opinionated guide on which AI Model to use in 2025.

The new way to build software.

The AI agent market map.

The Pivot to Product-Market Fit.

📰 News

Nvidia debuts Groot N1, a model for humanoid robotics.

Stability AI’s new AI model turns photos into 3D scenes.

Google released Gemma 3, the most powerful AI model you can run on one GPU.

Nvidia announces Blackwell Ultra and Rubin AI chips.

💸 Fundraising

Analytics company Dataminr secures $85M.

Prezent raises $20M to build AI for slide decks.

GrubMarket raises $50M to build AI for the $1 trillion food-distribution industry.

Virtual desktop startup Nerdio raises $500M.

Fail(St)ory

The Death of ANS

Last week, Flipkart shut down ANS Commerce, a SaaS platform focused on direct-to-consumer (D2C) brands, leaving over 200 employees without jobs.

Acquired in 2022 for $40 million, ANS Commerce was meant to help brands build and manage their own online stores, providing everything from digital storefronts to fulfillment solutions. Its shutdown raises doubts about whether independent SaaS providers can survive as big players like Flipkart and Amazon keep bringing these services in-house.

What Was ANS Commerce:

ANS Commerce aimed to make e-commerce simpler for brands by offering a full-stack solution. Instead of relying solely on major marketplaces, companies could use ANS to manage their own brand stores while still integrating with platforms like Flipkart and Amazon. Their services included:

• Brandstore Technology: Setting up and managing online storefronts.

• Performance Marketing: Driving traffic and conversions through digital advertising.

• Marketplace Management: Helping brands sell efficiently on multiple e-commerce platforms.

• Warehousing & Fulfillment: Handling logistics, inventory, and order processing.

At its peak, ANS Commerce worked with over 100 brands and had secured $2.2 million in pre-Series A funding in 2021.

The Numbers:

•  📅 Founded: In April 2017

• 💰 Acquired by Flipkart: 2022, for $40M

• 🏢 Brand Partnerships: Worked with 100+ brands

• 👷 Employees: Over 200 people

Reasons for Failure: 

• Strategic Shift by Flipkart: At the time of acquisition, ANS Commerce fit well into Flipkart’s vision of empowering brands with better e-commerce tools. But as competition with Amazon and Reliance intensified, Flipkart reassessed its priorities. Rather than supporting third-party SaaS platforms, it seems to be focusing on a more controlled, in-house approach to e-commerce.

• Financial Struggles: ANS Commerce was making more money than before, but it was also losing even more. In the last year, it generated about $6.5 million in revenue but lost $9 million in costs. This meant it was burning cash instead of getting closer to making a profit. For Flipkart, keeping a business that continued to lose money with no clear turnaround strategy didn’t make sense.

• The Broader SaaS Challenge in E-Commerce: Third-party SaaS solutions in e-commerce are increasingly at risk as major platforms consolidate power. The more marketplaces like Flipkart and Amazon build their own end-to-end solutions, the harder it becomes for independent enablers to justify their existence. If brands can get similar services directly from Flipkart or Amazon, why would they need a separate provider?

• Post-Acquisition Integration Issues: Many startup acquisitions fail not because the idea was bad, but because integrating a smaller company into a large corporate structure is complex. ANS Commerce had a unique culture and way of operating, which may not have meshed well with Flipkart’s broader ecosystem.

Why It Matters: 

• It highlights the risk of depending on acquisitions. If your startup is acquired, that doesn’t mean long-term survival—large companies cut non-core businesses all the time.

• SaaS for e-commerce brands is becoming a tougher market. If marketplaces build similar tools, third-party enablers lose their edge.

• Growing revenue isn’t enough—profitability matters. ANS Commerce was making more money, but its losses kept growing. If your business model can’t sustain itself, even rapid growth won’t save you.

Trend

Vibe Coding

Last week, Y Combinator’s CEO Garry Tan made a bold statement: "Vibe coding is the future." The term, originally coined by OpenAI co-founder Andrej Karpathy, refers to a new way of coding where developers rely almost entirely on AI tools to write, debug, and manage their code—without fully understanding or even reading it. Instead of carefully structuring logic, vibe coders just describe what they want and let the AI handle the rest.

This approach, naturally, has sparked intense debate. Some see it as the next step in making software development more accessible, while others believe it’s a dangerous shortcut that will lead to unmanageable codebases and unreliable software. So, is vibe coding the future, or just another passing trend?

Why It Matters:

• A new way to build startups – AI-powered coding is allowing startups to stay leaner than ever. Tan claims companies are reaching $10M in revenue with fewer than 10 employees—something unheard of before.

• A shift in developer roles – If AI can handle 95% of a codebase, as Tan suggests, traditional programming skills may become less critical. This could lower the barrier to entry for non-developers but also devalue specialized software engineering expertise.

• Not without risks – Developers are warning that overreliance on AI for coding could lead to more bugs, security issues, and projects that become unmaintainable.

What exactly is Vibe Coding?

The phrase "vibe coding" first appeared in a tweet by Karpathy, where he described the experience of using AI-assisted coding tools like Cursor Composer.

Instead of manually tweaking code, vibe coders simply tell the AI what they want—whether it’s reducing padding on a sidebar or fixing an error—and accept whatever solution the model generates, often without reviewing the changes in detail.

Y Combinator’s CEO Garry Tan took this concept further in a recent video, claiming that vibe coding is already reshaping how startups build software.

According to Tan:

• Startups are becoming dramatically more efficient, with small teams able to ship full-scale products faster than ever.

• AI-powered coding allows even those without a deep technical background to create software and launch profitable businesses.

• The vast majority of AI-first startups at YC are already coding this way. 25% of the current YC batch have codebases that are almost completely built by AI.

• Companies that don’t adapt may risk being left behind.

The Developer Backlash

Not surprisingly, many developers aren’t thrilled about this vision of the future. Some of the biggest concerns include:

• Loss of control – Developers argue that trusting AI-generated code without understanding it leads to unpredictable, difficult-to-debug projects.

• Short-term productivity vs. long-term problems – While AI can quickly generate functional code, maintaining and scaling that code over time is another challenge entirely.

• Poor debugging experience – Since vibe coding relies on blindly accepting AI-generated changes, fixing issues becomes a nightmare when things break.

• A growing reliance on specialists – The rise of vibe coding has already created a new industry: marketplaces where vibe coders can hire experts to fix their AI-generated messes.

I think this tweet sums up the feeling among many developers.

So, is Vibe Coding Real?

Yes and no. For small projects, personal tools, or MVPs, AI-assisted coding can be incredibly powerful. If your goal is to build something quickly and iterate fast, vibe coding might be the most efficient path. But when it comes to large-scale systems, AI still struggles to handle complex architectures, deeply interconnected codebases, and nuanced edge cases.

Will developers who don’t embrace vibe coding be "left behind"? Unlikely. While AI will continue to improve, software engineering isn’t just about writing code—it’s about understanding systems, solving problems, and making strategic decisions. Those skills will remain valuable, regardless of how much AI can automate.

Personally, I see vibe coding as another step in the evolution of software development. I’ve relied on no-code tools to build projects in the past, and if I were starting today, I’d probably be experimenting with vibe coding instead. It lowers the barrier to entry for many, but at some point, every project still needs structure, expertise, and real engineering.

Help Me Improve Failory

That's all of this edition.

Cheers,

Nico