Hey - It’s Nico.

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

If you only have one, here are the 3 most important things:

• StreamElements, a startup building creator tools, is about to shut down — learn why below

• The Dark Marketplace: why Vertical AI is coming for commerce

• Google’s I/O 2026 event just finished — learn every important announcement below

A huge thanks to today’s sponsor, SimpleClosure. When startups like StreamElements face an uncertain future, close cleanly and move forward with structured, expert-guided support.

Closing Well in Uncertain Times ^AD

StreamElements’ potential shutdown is a familiar startup story: a company can build for years, raise significant funding, support millions of users, and still reach a point where the next step is uncertain.

Across the creator economy and AI tooling space, founders are running into similar pressure: crowded markets, changing customer behavior, and harder paths to sustainable growth.

When shutdown becomes the right next step, the work is not over.

SimpleClosure helps founders bring structure to the process, from state filings and investor communications to final distributions, compliance, and asset sales. So you can close cleanly, responsibly, and with a clearer path through.

👉 Learn more

This Week In Startups

🔗 Resources

YC’s Request for Startups - Summer 2026

The Dark Marketplace: why Vertical AI is coming for commerce

After Automation: AI progress creates more work for humans, not less

Stuck between US LLC and EU? Honest comparison of 12 countries, real tax data * 

📰 News

OpenAI co-founder Andrej Karpathy joins Anthropic’s pre-training team

Elon Musk has lost his lawsuit against Sam Altman and OpenAI

OpenAI launches ChatGPT for personal finance, will let you connect bank accounts

Figma adds an AI assistant to its collaborative canvas

💸 Fundraising

Hark raises $700M Series A at a $6B valuation

Mercury raises $200M Series D funding

GridCARE raises $64M to squeeze more power out of grids

Commure, an AI platform for Healtchare, raises $70M in funding

* sponsored

Fail(St)ory

The Cost of Free

StreamElements was a creator-tools company built for livestreamers.

It had millions of creators, deep Twitch and YouTube roots, $111 million in funding, and a product people actually used. 

Then, in May 2026, the company entered shutdown mode while trying to find a buyer, with creators reportedly told to save their assets during a 30-day window.  

What Was StreamElements:

Launched in 2017 the company started with a simple pitch: give streamers the tools they need to run a professional live channel, for free.

That meant overlays, alerts, chatbots, tipping pages, loyalty tools, merch stores, analytics, moderation, and sponsorship infrastructure. The plumbing behind the show.

A streamer does not just press “go live.” They need alerts when someone subscribes, commands for chat, widgets for donations, sponsor graphics, and a setup that does not collapse the second the audience shows up.

StreamElements became one of the default options in that stack.

By 2019, more than 200,000 Twitch and YouTube channels were using it. Then the pandemic hit and livestreaming exploded. By 2021 more than 1.1 million creators were using the tool.

That same year, SoftBank led a $100 million round, bringing total funding to roughly $111 million.

So, they had the users, but how were they making money?

StreamElements kept the tools free, then made money through brand sponsorships. Brands paid to run campaigns with creators, creators got paid for the work, and StreamElements took a cut in the middle.

In theory, it was a good idea. Creators loved free tools. Brands wanted creator distribution. StreamElements sat in between.

However, free software still has bills.

By 2026, StreamElements said it was spending $300,000 per month just to keep its tools running. That did not include the heavier stuff around payroll, sales, support, and product.

So in January 2026, it launched Keep It Live, a crowdfunding-style campaign asking creators and viewers to help keep StreamElements free.

The campaign accidentally told the real story.

StreamElements said 23 million-plus creators had used its tools over the past decade, and 2 million-plus had gone live that month. The campaign had raised $1,700 from 75 supporters in the research snapshot.

Millions used it. Almost nobody paid for it.

By May 2026, GamesBeat reported that StreamElements was preparing to close its website and sunset its creator tools. Creators were reportedly told they had 30 days to log in and save their assets.

The company later said it was in “positive discussions with potential acquirers.”

The Numbers:

• 💰 Funding: $111M 

• 👥 Users: 23M+ creators served.

• 🧾 Costs: $300K monthly tool costs

• 📉 3 major layoff rounds between June 2022 and August 2023.

• 🧑‍💻 200+ employees at peak, down to about 72 listed on LinkedIn by May 2026.

Reasons for Failure: 

• The free-tools bargain broke. StreamElements built enormous reach, but most of that reach sat outside the paywall because there was no real paywall. That only works when sponsorship revenue is growing fast enough to subsidize the product.

• Sponsorships were too cyclical for the cost base. The company depended heavily on brand campaigns, and those budgets were the first thing to wobble when the ad market slowed. StreamElements said as much during its 2023 layoffs, pointing to weaker advertising demand and reduced budgets from existing clients. That is a bad place to be when your product costs are fixed and your revenue depends on marketers staying brave.
  

• The 2021 round sized the company for a boom that did not last. StreamElements raised $100 million when livestreaming, gaming, and creator monetization all looked like they would keep climbing forever. Then growth cooled, ad spend tightened, and the company had to cut staff three times in 14 months

• The creator product and the buyer product were not the same thing. Creators came for alerts, overlays, bots, and tipping tools. Brands came for measurable sponsorship performance. Those are two different products, two different sales motions, and two different definitions of success. StreamElements had creator adoption, but the business depended on advertiser conversion.

Why It Matters: 

• Creator love does not matter if the buyer is the advertiser.

• Boom-era funding can lock a startup into boom-era assumptions.

Trend

Google I/O 2026

Google just wrapped its I/O 2026 event with a ton of AI announcements.

Most of them point in the same direction: Google wants AI embedded across everything it already owns. Search, Gmail, Chrome, YouTube, Shopping, Android.

Gemini 3.5 Flash became the foundation for most of the launches. Spark handles long-running tasks in the background. Omni pushes AI deeper into editing workflows. Search gets more agentic. Shopping gets automated.

Here’s what matters and where this is going.

Why it Matters

• Google launched the operational model first: Gemini 3.5 Flash shipped before Pro because agents care more about reliability and speed across repeated actions than perfect reasoning on one big query.

• Google is quietly turning Search into a monitoring product: The shift is from “search this” to “keep watching this for me.” That changes how people discover products, content, and information online.

• Omni is a bigger threat to creative software than Veo was: Generating clips is one thing. Editing existing assets through conversation is much closer to actual creative work.

Gemini 3.5 Flash

The biggest launch from I/O was Gemini 3.5 Flash.

Google says it outperforms Gemini 3.1 Pro across several coding and agent benchmarks while running much faster. More importantly, Google immediately made it the default model behind AI Mode in Search.

Google clearly optimized Flash for constant execution across tools and workflows. Search queries, coding agents, background tasks, enterprise agents, shopping flows.

These systems constantly make API calls, retrieve information, compare outputs, and retry actions. Speed starts mattering a lot once that stack grows.

Spark

Spark feels much closer to OpenClaw-style agents than to a traditional assistant. Google is basically building a persistent cloud worker connected to Gmail, Calendar, Search, and eventually payments.

The examples were practical: reviewing inboxes, tracking internship openings, turning emails into task lists, scheduling calendar blocks, preparing recurring workflows, the kind of thing you would expect OpenClaw to be able to handle.

Spark runs on dedicated virtual machines inside Google Cloud, so it keeps running even when your devices are offline.

Google also said Spark will support third-party tools through MCP integrations later on. The rollout starts with trusted testers and AI Ultra subscribers in the U.S.

Daily Brief

Daily Brief organizes your morning inside Gemini: inbox updates, schedule context, tasks, reminders, priorities.

It’s nothing groundbreaking technically, but the packaging is probably the important part. Google is wrapping agent behavior into an existing habit instead of asking users to build workflows manually. Open app in the morning, get context, move on.

That approach showed up across a lot of the I/O announcements. Less “build your own AI system,” more lightweight automation built directly into products people already use.

Omni

Google also introduced Gemini Omni, a multimodal model focused on video generation and editing.

The important part is that Omni works with existing assets, not just prompts. Users can combine text, images, audio, and video inputs, then edit outputs conversationally inside tools like Flow and YouTube Shorts.

Search and Commerce

Google also upgraded AI Mode in Search with Gemini 3.5 Flash and introduced information agents that continuously monitor topics for users.

The shift here is pretty straightforward: instead of repeatedly searching for the same things manually, users hand monitoring tasks to the system.

Google connected this directly into shopping too.

Universal Cart and the new Agent Payments Protocol let users define spending limits, product constraints, and purchase conditions while agents handle transactions automatically once those conditions are met.

Overall, most of the I/O announcements pointed in the same direction: AI systems that stay connected to ongoing workflows instead of isolated prompts.

Help Me Improve Failory

That's all for today’s edition.

Cheers,

Nico