Why Flipboard just bought a two-person startup

Flipboard, the news magazine app that has pivoted hard toward the open social web, announced Wednesday that it’s acquiring Graze, a Portland-based startup that lets people build, customize, and monetize feeds on Bluesky.

It’s a small deal — Graze raised just $1 million at the start of 2025 and has only two employees. But the acquisition says a lot about where Flipboard CEO Mike McCue thinks the social internet is heading.

Graze has been live for 21 months. In that time, its feeds have pushed out over 41 billion posts to roughly 12 million users, according to the company. Today, more than 7,000 feeds run on its platform, and about 60% of that traffic is monetized.

How Graze makes money without tracking you

The core idea is simple: curators build feeds around a topic or community, and brands buy ad placements in those feeds. The curator gets to approve which ads run. The revenue is split 70/30 in favor of the feed creator.

What makes this different from the ads you see on Facebook or Google is the absence of cross-site tracking. No following users around the web. No surveillance-style profiling.

Instead, ads are contextual. A brand that wants to reach game developers buys a spot in a feed run by game developers. The audience is the context.

McCue told TechCrunch that the adtech “doesn’t have to surveil anybody, and it doesn’t have to violate people’s privacy.” He added: “The key thing is that all the economics aren’t just going to one player. The ad revenue is being shared with the feed builder.”

From Surf app collaboration to full acquisition

Flipboard had already been using Graze’s technology inside its new app, Surf, which is designed as a browser for the open social web. That collaboration revealed something: the two companies were building toward the same vision, but duplicating each other’s work.

Graze co-founder and CEO Devin Gaffney realized his startup could move faster with a partner that already had the advertising infrastructure — and the brand relationships — that Graze lacked.

Post-acquisition, Flipboard will keep using Graze’s tech and let users build their own feeds. That capability will expand beyond Bluesky to everything running on the AT Protocol, including Flipboard’s own app.

What is the AT Protocol, anyway?

The AT Protocol is an open technical standard that lets different apps share the same underlying social network. Think of it as a common language for social platforms. Bluesky is built on it, and so is Surf. The idea is that you shouldn’t be locked into one company’s app to access your social graph.

The human side of feed curation

McCue made an interesting point in the interview about how users actually think about these feeds.

“They don’t think of it as, like, ‘I’m joining Bluesky’ or ‘I’m joining the social web, the open social web.’ They just think I’m joining The Tea because ‘I care about women’s sports,'” he said, referencing a Surf community feed called Bet on Her.

That’s the bet Flipboard is making: people will follow interests, not platforms. And if that’s true, then the people who curate those interest-based feeds deserve a cut of the ad money.

What this means for the open social web

The acquisition is another signal that the open social web is moving from a niche experiment to something with real economic structure. Flipboard has been investing heavily in this space, and Graze gives it a proven monetization layer.

For feed creators, the deal means more tools and, potentially, more revenue opportunities. For brands, it offers an alternative to the surveillance-based ad model that has dominated digital advertising for two decades.

Whether that alternative scales remains to be seen. But with Flipboard’s infrastructure behind it, Graze’s approach to contextual ads on open social web feeds just got a much bigger stage.

Terms of the deal were not disclosed.

Why “Prompting” Isn’t Enough Anymore

You’ve probably gotten good at asking ChatGPT or Claude for a quick email draft or a summary. But that’s not building a workforce — that’s just typing faster.

The real shift happening right now is bigger. People are moving from one-off prompts to creating AI employees: systems that take a task, run with it, and hand back finished work without you babysitting every step.

Sound like science fiction? It’s not. It’s already being done by small teams and solo operators who’ve figured out a simple loop: build, train, schedule, review. Here’s how that works in practice.

Step 1: Define the Job Before You Build the Bot

Before you write a single prompt, sit down and write a job description. What exactly do you want this AI employee to own? Not “help with marketing” — that’s too vague. Think: “Draft a weekly newsletter from my blog posts, pull key quotes, and format it for Mailchimp.”

A clear job scope is what separates a useful AI tool from a toy. Without it, you’ll spend hours tweaking outputs that never quite hit the mark.

Start with one repetitive task

Pick something you do every week that takes you 30 minutes or more. That’s your first AI hire. The goal isn’t to automate your whole business overnight — it’s to free up one chunk of your calendar.

Good candidates:

• Summarizing meeting notes into action items
• Generating social media posts from a blog URL
• Drafting client follow-up emails in your voice
• Creating weekly status reports from project data

Step 2: Train Your AI Employee With Your Standards

Here’s the part most people skip: training. You don’t just hand an AI a task and hope for the best. You teach it your style, your tone, your non-negotiables.

Start by feeding it examples. Show it three or four pieces of work you love — a great email you wrote, a newsletter that got compliments, a report that impressed a client. Tell the AI: “This is the standard. Match it.”

Then, correct it. When the output misses, don’t just accept it. Tell the AI what’s wrong and what you want instead. Over time, it learns. This is the training loop, and it’s the difference between generic AI output and something that sounds like you.

Create a style guide for your AI

If you want consistency, write a short document your AI can reference. Include your brand voice, preferred phrases, things to avoid, and formatting rules. Then paste it into your custom instructions or system prompt.

This isn’t a one-time thing either. Your standards evolve, so your AI’s instructions should too. Review and update them monthly.

Step 3: Schedule Your AI to Work Autonomously

Once your AI employee can produce good work, it’s time to set it loose. This is where scheduling comes in.

Tools like Make, Zapier, or n8n let you connect your AI to your other apps and trigger it on a schedule. For example, every Monday at 9 AM, your AI could:

• Pull last week’s blog posts from your CMS
• Generate a newsletter draft
• Send it to you for approval

That’s autonomy with a safety net. You’re not fully hands-off yet — you’re reviewing, not creating.

From review to full delegation

As your AI gets better, you can widen the scope. Move from “send me a draft” to “publish it, and notify me if anything looks off.” The trust builds gradually, just like with a human employee.

Some teams even let their AI handle customer service triage, sorting inquiries and drafting responses for a human to approve. The key is to start with low-risk tasks and scale up as your confidence grows.

Step 4: Measure, Adjust, Repeat

An AI employee isn’t a set-it-and-forget-it thing. You need to check the quality regularly.

Set a weekly review session — 15 minutes, max. Look at what your AI produced, spot any errors or drift, and tweak the instructions. Over time, the error rate drops, and you can delegate more.

This is also where you decide what to automate next. Once one task is running smoothly, pick the next repetitive chore and build another AI employee. Before long, you’ve got a small team working around the clock — and you’re only stepping in for the final checks.

Common Mistakes to Avoid When Building AI Employees

You’ll save yourself a lot of frustration if you avoid these traps:

• Expecting perfection immediately. Your first outputs will be rough. That’s normal. Training fixes it.
• Skipping the style guide. Without it, your AI will sound like a robot — or worse, like every other business using the same tool.
• Automating before you understand the task. If you can’t do the task yourself in 10 minutes, you can’t teach an AI to do it well.
• Not reviewing. Full autonomy without oversight is how mistakes snowball. Keep a human in the loop for now.

Building AI employees isn’t about replacing your team — it’s about giving yourself back time. Start small, train hard, and let the machines handle the repetitive stuff while you focus on the work only you can do.

For more on getting started, check out our guide on AI workflow automation and how to train AI for business tasks. If you’re new to the basics, our AI for beginners primer will get you up to speed fast.

Editing Reels has always been a chore. That might be over now.

If you’ve ever spent 20 minutes trimming a single Reels clip down to the good three seconds, you know the pain. Instagram is finally doing something about it.

The company announced Tuesday in a Threads post that it’s rolling out Instagram First Draft, a new editing feature that automatically trims your selected video clips and strips out the awkward pauses. The result? A rough cut of your best moments, ready for you to refine.

No more scrubbing through footage frame by frame. The feature does the boring part for you.

How First Draft works

Here’s the gist: you pick the clips you want to use, and Instagram’s algorithm goes to work. It identifies the action, cuts the dead air, and stitches together a first pass. You still get full control afterward — it’s a starting point, not a final product.

Think of it like a rough edit from an assistant editor. You get the bones, then you add the flair.

What it cuts (and what it keeps)

• Pauses — those awkward beats where nobody’s talking or moving.
• Dead frames — shaky transitions or moments where the camera just sits there.
• Best moments — the algorithm prioritizes action and energy, so your hook lands faster.

It’s not perfect, and it won’t replace a skilled editor. But for quick, casual Reels — the kind most people post — it’s a massive time-saver.

Why this matters for creators

Editing is the most hated part of content creation. A 2023 survey from Adobe found that 71% of creators spend more time editing than actually filming. That’s a huge drain on creativity.

Instagram knows this. The platform has been losing ground to TikTok’s built-in editing tools, which have always been more forgiving. First Draft is a direct response — an attempt to make Reels feel less like homework.

It’s also a smart move for retention. The faster someone can post, the more they’ll post. And more posts mean more time on the app.

When you’ll get it

The feature is rolling out now, though Instagram hasn’t given a precise timeline for global availability. As with most Instagram features, expect a gradual rollout — some users will see it this week, others in a month.

To check if you have it, open the Reels editor and look for a “First Draft” option in the toolbar. If it’s not there yet, it’s coming.

What this means for your workflow

If you’re a casual poster, this could cut your editing time in half. If you’re a professional, it’s a decent starting point for quick cuts before you move to Premiere Pro or CapCut for fine-tuning.

Either way, the bar for entry just got lower. You no longer need to be a video editing whiz to make a Reel that doesn’t feel amateurish.

And honestly? That’s a win for everyone who’s ever stared at a 10-minute clip and thought, “Where do I even start?”

Now you start with First Draft. The rest is up to you.