I used Grok Bot this week on Stack Snacks. The moment it clicked was not a feature tour. I asked it to clip one of my old YouTube videos into shorts and just give me the files back. At the same time I was chatting with my Chief of Staff bot about how to organize the September project. I was not bouncing between tabs, re-explaining the week to a blank chat. It felt like I had a team. That has been fun to experiment with, and a joy to use.

I would keep it. My Grok/X/Cursor whatever subscription I am on I am going to keep because this was a lot of fun.

What it actually does

Grok Bot is not Grok the chatbot. You create agents with a job, you put them in a channel, and they share a computer. They can fetch a video, cut files, look at a board, and keep talking to you while another bot handles the calendar.

I stayed on desktop. I have not taken this on my phone.

The setup that worked: a Chief of Staff to keep the other bots honest, a content bot for this list, and a Video Prep channel. The job was four Friday videos in September, same piece as the Tuesday email, no news recaps. When I said I did not want another Opus 4.8 video, they swapped that row to Cursor and punched the hooks the same night. Separately, I pointed it at an old Gemini Spark walkthrough and got three vertical shorts back: the roast, the spreadsheet payoff, and the $20 verdict.

That is the product. A room that can clip and plan at the same time.

Where it earned its keep

Two jobs at once, without me becoming the glue. The shorts came back as files I could hold. The calendar came back as titles, cold opens, and Hub rows. I did not have to paste a brief from one window into another.

It produces things I can ship. Not "here is a strategy." Files. Lines I would actually say on camera. A tweet I wanted to post. This email.

It is easy enough that I opened it again the next day. I did not need a weekend course. I named a job, named a constraint, and let them work. The joy is not a strategy, but it is why this lasted more than one night.

Where it failed

Desktop only. If the pitch is that it runs while you walk around, that was not my week.

You still have to name the job. "Come up with video ideas" is how you get a recap nobody asked for. "Four videos, one original a week, skip Hermes, clip this URL" is how you get a month and three files.

Credits are not free just because you have not hit the wall. I am on X Premium+. I have not run out. I have been watching what eats the most. Clipping a whole video is the kind of job I would not fire twice in a day just to see.

How I would start Monday

Two bots. One channel. One project you already have.

Give one bot a file job (clip this, clean that, fetch the page). Give the other the project (what are we shipping, what are we not). Stay in the room for decisions only you can make.

Then do the thing that got me: ask for the files back, and keep talking about the calendar in the same breath.

Prompt stack: briefing the room

Steal this. I used it this week.

You are my Chief of Staff. There is a content bot in this channel with you.

Job 1: organize September. Four Friday videos. Same piece as the Tuesday newsletter. No news recaps. Skip anything that is not ready to film.

Job 2: clip this YouTube URL into three vertical shorts and give me the files. Do not make me download anything by hand.

If I change the brief, update the board. Do not start a second calendar. Tell me which job is eating credits if it starts to look expensive.

When to use it: the first message in a new channel

When not to: a one-off "what should I post" with no file and no board

Model I used: Grok Bot, on X Premium+

Tweak: swap Job 2 for whatever file you have been meaning to finish (a transcript, a sheet, a clip)

What it costs

I am on X Premium+. Grok Bot comes with that plan. I have not hit the weekly usage wall. Extra use can go on-demand, so I am watching the heavy jobs. Compared to paying another chatbot seat and doing the clip + the calendar by hand, this is the first time the seat felt like a team. I would not upgrade just to try the trailer. Use the plan you already have. See if two bots earn the week.

Current status: Keeping it.

This is how I am running Stack Snacks in September.

If you want the short version: ask it to clip an old video and hand you the files, and in the same sitting talk to a Chief of Staff bot about the actual project. If that feels like a team, keep it. If you have to babysit every sentence, cancel.

If this was useful, forward it to one person who is about to buy another chatbot seat.

See you next Tuesday.

This isn’t a ranking of the “best” AI tools. It’s an honest look at what I’m currently paying for, why each tool has earned a place in my stack, and which subscriptions I’m reconsidering.

There are a lot of excellent AI products available, but paying for all of them quickly gets expensive. Here’s where my AI budget is going right now.

Zo Computer: The Center of My AI Stack

Zo has become the first thing I open whenever I sit down at a computer.

It gives me access to my files, lets me switch between different AI models, and helps me build almost anything I need. Instead of being another chatbot subscription, it feels more like the operating system for my AI workflows.

It is now the number-one tool in my stack. I can also connect my ChatGPT subscription, which helps me get more value from the tools I’m already paying for.

Current status: Essential. I’m keeping it.

ChatGPT Plus: My General-Purpose Workhorse

ChatGPT has been my everyday AI tool for the longest.

I still use it for most of my general-purpose work: thinking through ideas, writing, research, coding, and building projects. The addition of Codex and the newer ChatGPT Work features has made the subscription even more useful.

I also like being able to use my ChatGPT subscription with tools such as Zo Computer. That flexibility makes the $20 subscription feel like it stretches much further than ChatGPT alone.

Current status: Still one of the easiest subscriptions to justify.

Claude: The Subscription I’m Testing Life Without

The Claude family of models is incredible, but I’m planning to cancel my subscription and see how much I actually miss it.

The biggest issue isn’t model quality—it’s flexibility. I can’t get as much value from my Claude subscription inside other products, and not having access to Fable, even for a few prompts per day, makes the subscription less compelling for me.

My plan is to use Claude through the API whenever I specifically want a Claude model. Then I can compare the actual API cost against the monthly subscription.

This is also a useful test: do I genuinely need the Claude subscription, or am I keeping it because I’m used to having it?

Current status: Likely canceling and seeing how work is without it.

Gemini: My Least-Used Model but Possibly My Best Bundle

Gemini is the strangest subscription on this list.

It is probably the AI model I use the least, but the overall subscription might provide some of the best value. Google bundles several benefits into its AI plan, including additional Google Drive storage and other Google ecosystem perks.

I’m probably not taking advantage of everything included, but I already use enough Google products that the subscription is easier to justify—even if Gemini itself isn’t my first choice for most AI tasks.

Current status: Keeping it for the bundle, not necessarily the model.

Grok: The Unexpected Comeback

I’m surprised Grok made this list.

The early versions were interesting, but I never felt they were strong enough to compete with the major AI models. Slowly but surely, though, Grok kept improving.

Starting around Grok 4.5, the combination of model quality, inexpensive API access, and integration with X became much more compelling. You can also authenticate an X account with agents such as Hermes Agent, which opens up some genuinely useful workflows.

I’ve been using Grok more and more, especially for X-related research, content, and automated workflows. It has gone from something I occasionally tested to a model I can see becoming a regular part of my stack.

Current status: Increasingly useful—and one to watch.

What I’m Learning From This

The best AI subscription is no longer determined only by which company has the best model.

I’m increasingly looking at three things:

1. Flexibility: Can I use the subscription outside the company’s main chatbot?

2. Ecosystem value: Does it include storage, credits, tools, or other services I already need?

3. Actual usage: Am I paying because I use it, or because I’m afraid I might need it?

My current strategy is to keep subscriptions that act as platforms, use APIs for models I only need occasionally, and regularly test whether I can live without everything else.

The AI subscription pile can grow quickly. The harder—and more useful—question is which tools are actually earning their place.

Scheduled tasks in ChatGPT aren’t new — but OpenAI just gave them a fresh coat of polish and a couple of features that make them genuinely worth a second look. There’s now a dedicated Schedules tab right in the sidebar, and once you start wiring these up to your actual workflow, things get interesting fast.

Here’s what I’ve got running, what’s actually useful, and where it still feels a little half-baked.

What Are Scheduled Tasks?

The idea is simple: you give ChatGPT a prompt and a schedule (morning, afternoon, evening, or night — daily, weekly, whatever), and it runs automatically and delivers the result to you. It shows up on your phone too, which is honestly half the appeal. No hunting, no remembering to ask.

My Setup: The Tasks I Actually Keep

1. The Daily AI Briefing

Every morning, ChatGPT goes out and pulls together new AI tools, big industry moves, and a few content ideas. Since I’m a personal finance nerd, I also had it flag money-saving opportunities.

The verdict: The industry highlights are legitimately good — today’s hit on the G7 Summit and agent security, stuff I’d otherwise have to go dig for. The money-saving picks, though? They kind of feel like ads. Mixed bag. But the “suggested action for today” section (make a Short, draft an offer, test a workflow) is a nice nudge to actually do something with the info.

2. The Daily (or Weekly) Review

This one’s my sleeper favorite. The prompt is basically:

“Tell me a summary of everything I did in the past day. What have you observed? What am I thinking about? Give me your honest thoughts.”

I’ve actually been running this as a weekly review, and if ChatGPT is your daily driver, the insights it hands back are surprisingly interesting. It’s like getting a mirror held up to your own week. Try it for seven days — I’m telling you, it’s worth it.

3. The Fun One: World Cup Recaps

One of OpenAI’s default suggestions was a World Cup recap, and honestly? With everything going on in the tournament, getting a scheduled roundup of scores was kind of cool. Not everything has to be productivity.

Where It Gets Actually Powerful: Connecting Apps (MCPs)

Here’s the update that matters. You can now connect your apps — what ChatGPT calls “apps,” which are essentially MCPs — directly into scheduled tasks.

So I set one up: every Sunday night, summarize everything in my Todoist inbox so I can plan my week. ChatGPT connected to my task list, pulled the inbox, and now that review lands automatically before Monday hits.

That’s where this stops being a gimmick and starts being a real workflow tool. Calendar digests, task reviews, inbox summaries — once your tools are plugged in, you can build some genuinely custom automations with a single prompt.

The Honest Verdict

Nothing here is super duper new — scheduled tasks have existed for a while, and the early version honestly wasn’t that great. But they’ve come a long way:

• The good: Dedicated sidebar tab, mobile notifications, app/MCP connections, and briefings that actually save you hunting time.

• The meh: Some of the auto-generated content (looking at you, money-saving tips) feels generic or ad-adjacent.

• The move: Start with two or three tasks max. A morning briefing, a weekly review, and one task connected to a tool you actually use. Don’t over-build it — I only run three or four myself, and that’s plenty.

If ChatGPT is already part of your daily routine, this is a low-effort way to make it work for you in the background instead of waiting for you to show up and ask.

Do you use scheduled tasks? Got a better setup? Let me know — I’d love to hear what workflows people are building.

Thanks for stopping by. See everybody next time. 🍿

Notion has been moving deeper into AI agents, and I wanted to answer the practical question:

Are Notion’s custom AI agents actually useful, or are they just another expensive AI add-on?

I use Notion a lot already. It is where I store documents, databases, content ideas, and a bunch of working notes. So the pitch is pretty appealing: instead of copying information out of Notion, pasting it into ChatGPT, then pasting the result back into Notion, what if an agent could work directly inside the system?

That is the promise.

So I built one.

The Goal

The agent I wanted was simple: take a rough video idea and turn it into a more complete content brief.

For Stack Snacks, that means things like:

• A clearer hook

• Possible titles

• A video description

• Keyword ideas

• Relevant news or context

• A saved brief inside a Notion database

• Maybe a follow-up task in Todoist

Basically, I wanted to see if a Notion agent could act like a mini content teammate.

Not a magic robot. Not a full creative director. Just a helper that can take a messy idea and move it one step closer to something usable.

Building the Agent

Quick clarification first, because Notion has two things called “agents” and the names are confusing.

There is the personal Notion Agent — the assistant in the bubble at the bottom of your screen. It works on demand, only when you prompt it, and it does not burn credits.

Then there are Custom Agents — the ones you configure to run on their own, triggered by schedules or database changes. These are the autonomous ones, and these are the ones that cost credits. That distinction matters a lot once we get to pricing, so keep it in mind.

What I built is a Custom Agent.

I started inside Notion’s agent area, where you can see existing agents and browse the marketplace. One of the more useful pieces I found was a Notion Agent Builder. Instead of writing the whole agent prompt from scratch, I used that builder to help create the first version of the instructions.

The prompt was roughly:

I want to build an agent that helps me take raw video ideas and turn them into a complete video brief.

The builder came back with proposed instructions and a few clarifying questions. From there, I created a new custom agent and started adding the pieces I wanted:

• Save new briefs into a Notion database

• Research useful keywords

• Check for relevant news articles

• Pull in live information where possible

• Draft an email when finished

• Create a related task in Todoist

That was the ideal version.

The actual version took a little more fiddling.

Where It Worked

The core workflow worked better than I expected.

I gave the agent a rough idea for a video about making AI videos with Gemini. It searched the web, started building a brief, and saved output into the database.

The result was not perfect, but it was useful. It pulled together a rough angle, gave some terminology guidance, and created a starting point I could edit. For example, it flagged that I shouldn’t conflate Veo — Google’s established video model — with Gemini Omni, the newer model that’s now replacing Veo inside the Gemini app. Those are genuinely two different things in mid-transition, and getting them mixed up in a video is exactly the kind of small error an AI-savvy audience notices. That is the kind of catch that can save you during planning.

It also successfully created a Todoist task after I connected Todoist through MCP.

That part is interesting. Notion agents can connect to external tools through custom MCP servers, which means the agent does not have to stay trapped inside Notion. It can potentially touch tasks, emails, APIs, and other systems if you wire them up correctly.

That is where the product starts to make more sense.

If your team’s work already lives in Notion, and you want an agent to operate on that shared context, this could become useful fast.

Where It Got Messy

The messy parts are exactly what you should expect if you have built any agent workflow before.

The agent created multiple database pages when I only wanted one.

It tried to draft an email, but I had not clearly told it who to send the email to, so that failed. Honestly, I was glad it stopped. That is better than an agent confidently sending something to the wrong person.

The Todoist MCP connection also had some issues before it worked. It eventually created the task, but it was not a perfectly smooth setup.

This is the real lesson: agents are not “set it and forget it” on the first try.

You have to watch what they do, tighten the instructions, remove risky actions, and build in constraints. In my case, I removed the email step because it added more risk and confusion than value for this specific workflow.

That is probably the right way to think about these tools:

Start with one safe action. Make it reliable. Then add more.

The Credit Problem

The biggest downside is cost — and this is the part where the details matter, because they changed recently.

First, Custom Agents are a Business/Enterprise feature. If you are on Free or Plus, you do not have them at all. Practically, that means the entry point is the Business plan at $20/user/month before you have run a single agent.

On top of that, Custom Agents run on Notion Credits, which are a separate add-on: $10 per 1,000 credits, pooled across your workspace, reset every month, no rollover. Notion ran a free exploration period through May 3, 2026, but billing started May 4 — so there is no free ride left.

How fast do credits burn? Each run costs anywhere from a fraction of a credit to a few dozen, depending on how much the agent reads, how many tools it touches, and which model it uses. For moderate use, most estimates land around $20–50/month on top of your Business seat.

That changes how I think about using it.

If I am already paying for Notion, ChatGPT, Claude, Gemini, or other AI tools, I do not want to accidentally create another $30 or $40 per month workflow unless it is clearly saving time.

The model picker helps. You can drop to a cheaper model for simpler tasks instead of running everything on the most expensive one. That is good. But it still means you need to pay attention to usage.

For a solo creator, I would not build twenty Notion agents right away. I would build one tight workflow and see if it actually earns its keep.

Who It’s For (and Who Should Skip It)

Notion Agents make the most sense for small teams or businesses already running work through Notion.

If you have shared databases, docs, project notes, meeting notes, content calendars, or client information in Notion, an agent sitting inside that context could be genuinely useful. Good use cases:

• Turning meeting notes into project tasks

• Summarizing client research into a one-page brief

• Creating content briefs from raw ideas

• Generating weekly updates from team databases

• Checking a knowledge base before answering internal questions

The key is that Notion already has the data. If the agent has to leave Notion, search around, talk to other apps, and then come back, the workflow gets more fragile. But if the work is already sitting in Notion, the value is much clearer.

If you are a solo user who only uses Notion as a personal notes app, this may be overkill — especially now that the free period is over. You can get a lot of the same value by using Notion’s MCP inside another AI tool, or by copying a page into ChatGPT or Claude when you need help. Not as elegant, but cheaper and simpler.

The real risk is building a complicated agent because it feels cool, not because it saves you time. I would avoid workflows that:

• Send emails automatically before you trust the system

• Create lots of database entries without a review step

• Touch multiple external tools at once

• Use expensive models for simple formatting tasks

• Run repeatedly without clear limits

Those are the places where small mistakes turn into annoying cleanup.

The Lean Version I’d Actually Build

If I started over, I would resist every urge to make it clever. Here is the actual instruction I would give the agent builder:

You help me turn a rough video idea into a single content brief.

When I give you an idea:
1. Ask me 2–3 clarifying questions max (audience, angle, what I already know).
2. Do light web research to check terminology and find one or two
   relevant, current references.
3. Create ONE new page in the [Content Briefs] database with these fields:
   - Hook (one sentence)
   - 3 working titles
   - Draft description (3–4 sentences)
   - 5 keyword ideas
   - Notes / things to fact-check
4. Stop there. Do not send emails. Do not create tasks in other tools yet.

Use a cheaper model unless I tell you the research quality isn't good enough.

That is it. One input, one output, one place. No external side effects until the brief format is reliable.

Once that is boring and dependable, then I would add the Todoist task. Then maybe a draft email — but only once I trust it. Start with one safe action, make it reliable, then earn the next one.

Set a monthly credit cap while you are at it, so a runaway agent cannot quietly drain your balance.

My Take

I like the direction. Working directly inside Notion is genuinely useful, the agent builder lowers the barrier, and MCP support means these agents do not have to stay trapped inside Notion.

But it is not a must-use for everyone. For solo creators, it is a “test one workflow with a tight credit cap” product. For small teams already running work through Notion, it is more compelling — the data is already there, so the agent has something real to act on.

Either way, the bar is not “can it do something impressive once?”

It is: can it quietly make your normal workflow easier every week?

Build one boring agent. Watch what breaks. That will tell you more than any demo.

I paid for Google’s Gemini Ultra plan ($100/month) to get early access to Gemini Spark, and I came away with a split opinion: this is one of the most capable hosted agents I’ve used, and I still can’t tell most people to buy it yet.

Spark is closer to “a junior operator on call” than anything else I’ve tried. It takes longer-running tasks, uses more compute, and works across your files instead of just chatting about them. The catch is that all of that power stops at the edge of Google’s ecosystem—and at $100/month, that edge is hard to ignore.

Here’s what stood out after hands-on testing.

What Spark does really well

Docs and Sheets workflows are fast and seamless. My first test was simple: can Spark actually do work inside the Google suite? It passed. Creating docs, spinning up spreadsheets, and moving between them was quick and genuinely impressive. If your day is “outline a plan, build a sheet, summarize a thing,” Spark delivers value on day one.

Calendar feels native, not bolted on. Scheduling-style tasks were smooth, which tracks given how deep Calendar already sits in the Gemini experience. It read like a real extension of what Google does well, not a demo feature.

It personalizes using everything Google already knows about you. When I asked it to put together a plan for my toddlers, it already knew how many kids I have and their ages—no restating. That’s equal parts creepy and useful. Creepy because you feel the surveillance-adjacent reality of living inside one company’s ecosystem. Useful because the agent finally stops acting like a goldfish. If you’ve been waiting for AI tools to remember context, this is a real glimpse of it.

Where it gets interesting: longer-running tasks

Auditing my YouTube channel was a standout. I had Spark review my channel and write up what’s working, what isn’t, and what to do next. It’s strong at pulling info from the web, synthesizing it, and handing back something actionable. The remote-browser view—where you watch it work—makes it feel less like magic and more like an operator you can supervise.

The chore automation is the real tell. I set up a recurring workflow to watch my email for rewards-card updates and log new offers to a master sheet. Yes, an app already exists for that. The point isn’t novelty—it’s that Spark collapses the setup cost. There was no integration to wire up, no automation tool to learn. I described the chore in plain English and it just ran. That’s the upside of this whole category: quietly removing dozens of micro-tasks from your week without the usual config tax.

The drawbacks—and why $100 is a hard sell

The integration gaps are oddly glaring. There’s no Google Tasks or Google Keep support. Those are already Google products, and they’re exactly the lightweight inputs that would make an agent feel sticky. Leaving them out makes the ecosystem story feel half-finished.

There’s no real extensibility, and that’s the dealbreaker. Spark doesn’t connect to anything outside Google’s walls. No way to plug in third-party tools or data sources—nothing like MCP (Model Context Protocol), which is fast becoming the standard for letting agents securely reach other systems. For comparison, the hosted agents I’ve used from Anthropic and OpenAI lean hard into exactly this kind of openness, and it’s what makes them useful for real cross-system work. Spark is powerful right up until you try to extend it past Google, and then it hits a wall.

So is Gemini Spark worth $100/month?

For most people, not yet. If your work is genuinely Google-suite-heavy—docs, sheets, calendar, web research—and Spark can reliably offload it, you can probably justify the cost today. For everyone else, it’s still early.

But I want to be clear about which way this leans: I used Spark more than I expected to, and it’s one of the most promising Gemini implementations I’ve touched. The problem isn’t capability. It’s that the price is running ahead of the openness. The day Google adds real extensibility, I’d resubscribe without thinking about it.

What would flip it to an instant yes:

1. Native Tasks and Keep support, at minimum

2. A first-class integration layer (MCP-style extensibility)

3. Third-party tool coverage beyond Google’s core suite

4. A lower tier for light-agent use, so the price matches the current capability

Until then, Spark is a compelling preview of what hosted agents can be—just not the everyday workhorse it’s trying to become.

1. GPT Image 2.0 — The jump I wasn’t expecting

For anything image-related, I’ve been living in Nano Banana and that family of models. Cheap, fast, good enough.

GPT Image 2.0 feels like a different class of tool. I gave it my website and asked for a redesign — one shot, no chain-of-prompts. What came back was genuinely usable. Not “AI slop you then fix,” but something a designer would actually start from.

What stood out:

• Text in images finally works — headlines render cleanly.

• Layout intuition is real — it understood hierarchy without me spelling it out.

• Style consistency across variants — three options felt like a set, not three random generations.

If you’ve been ignoring OpenAI for image work because Nano Banana was “good enough,” this is worth reopening the tab for.

2. Claude Design — Not a designer killer, but a great thinking surface

I don’t think it replaces a designer yet. But as a wireframing and prototyping canvas, it’s solid. Getting rough ideas out of my head and onto something I can click around with — that’s the sweet spot.

Where it struggled: full redesigns were mixed, and credits disappeared fast, which cut my testing short.

Where it shined: quick prototypes and early-stage exploration when I don’t know what I want yet.

Not a stack-changer — but it’s earned a slot.

3. Capacities MCP Server — Quiet, useful, exactly what MCP is good at

Capacities is a note-taking app I really like, and they just shipped their MCP server.

Here’s the pattern I keep finding with MCP: it’s magic when you want to push info from your AI chat into a tool. Capture an idea mid-conversation, drop it into the right object in Capacities, done.

Less flashy than the two above — but that’s kind of the point. It just works, and now my notes app is one less place I have to manually sync.

Key Takeaways

• GPT Image 2.0 — legitimate quality jump over Nano Banana.

• Claude Design — great for wireframes, not full design replacement yet.

• Capacities MCP — small but meaningful upgrade if you’re already in the app.

• MCP keeps proving its value as the glue between AI chats and your real tools.

Full GPT Image 2.0 redesign breakdown (plus prompts) coming to the Stack Snacks YouTube this week. 👉 [youtube.com/@stacksnacks]

P.S. — If you’ve tested GPT Image 2.0 on your own, hit reply and send me the output. Want to see how far this stretches.

Hey Snackers 👋

Anthropic dropped Claude Opus 4.7 last week — an upgrade to their flagship model — and this one’s worth a real look.

The headline: this is a coding model you can actually hand work off to. Not “hand it simple stuff and babysit it” — hand it hard, multi-step, hours-long tasks and walk away. At least, that’s the claim.

But there’s more going on under the hood than just better code. Better eyes, tighter instruction following, and a brand new approach to cybersecurity that’s quietly one of the most interesting things any AI lab has done this year.

Let’s break it down.

The Big Upgrade: Software Engineering

The main event with Opus 4.7 is coding. Anthropic says early testers were able to delegate their hardest, most supervision-heavy tasks to the new model with confidence — something they couldn’t reliably do with Opus 4.6.

What makes it different? Three things:

1. Self-verification. The model checks its own work before reporting back. It catches logical faults during planning and doesn’t just plow ahead hoping for the best.

2. Long-horizon consistency. Multi-step tasks that span many turns — the kind that used to degrade or lose the plot — now stay on track.

3. Benchmark results. State-of-the-art on SWE-bench (the standard software engineering benchmark), including multilingual and multimodal variants. On a 93-task coding benchmark, it improved resolution by 13% over Opus 4.6 — including 4 tasks that no previous Claude model could solve.

One early tester put it well: “Low-effort Opus 4.7 is roughly equivalent to medium-effort Opus 4.6.” Translation: it’s doing more with less thinking time.

4 More Things Worth Knowing

1. 3x Better Vision

Opus 4.7 accepts images up to 2,576 pixels on the long edge (~3.75 megapixels). That’s 3x the resolution of previous Claude models. If you’re using Claude for reading screenshots, extracting data from diagrams, or any computer-use agent work — this is a big deal. Heads up: higher-res images eat more tokens.

2. Instruction Following Got Too Good

This one’s interesting. Opus 4.7 takes instructions much more literally than previous models. Anthropic actually warns that prompts written for older versions might produce unexpected results — because older models would loosely interpret or skip steps. If you’re migrating from 4.6, re-tune your prompts.

3. Better Memory Across Sessions

The model got better at using file system-based memory, meaning it remembers context across multi-session projects. Less “let me re-explain everything” at the start of each chat.

4. Tastier Creative Output

Anthropic says the model produces higher-quality interfaces, slides, and docs. Less “AI-looking,” more polished. This matters if you’re using Claude for anything client-facing.,

Pricing & Availability

• Price: $5/M input tokens, $25/M output tokens (same as Opus 4.6)

• API string: claude-opus-4-7

• Available on: Claude.ai, API, Amazon Bedrock, Google Vertex AIv

• Tokenizer change: Updated — same text may use 1.0–1.35x more tokens than 4.6

Migration note: The tokenizer change means your costs could creep up even at the same per-token price. Anthropic recommends measuring real-world token usage on your workloads before fully switching over.

Also launched yesterday:

• xhigh effort level — a new setting between “high” and “max” for finer control over reasoning depth vs. speed

• Task budgets (public beta) — guide Claude’s token spending across longer runs

• /ultrareview in Claude Code — a dedicated review session that audits your code for bugs and design issues

• Auto mode for Claude Code Max — lets Claude make permission decisions on your behalf for longer, uninterrupted task runs

The Bigger Picture

Three trends converging here:

1. The “unsupervised coding” bar keeps rising. Six months ago, handing an AI a complex multi-file refactor and walking away was a fantasy. Now the top models are genuinely getting there. Opus 4.7’s self-verification isn’t a gimmick — it’s the feature that makes long-running agentic work practical.

2. Capability shaping is the new safety conversation. Forget content filters and guardrails. Anthropic is experimenting with making models structurally less capable in dangerous domains. That’s a fundamentally different approach than “add a warning label.” If it works, expect every lab to follow.

3. The model treadmill is exhausting — but the gaps matter. Yes, there’s a new model every week. But Opus 4.7 vs. 4.6 isn’t incremental — 13% better on hard coding tasks, 3x vision resolution, and a new safety architecture. If you’re building on Claude, this is a real upgrade.

Should You Care?

If you code with Claude daily: Yes. Full stop. The self-verification and long-horizon improvements are real. Start with your hardest tasks and see if the claims hold up. Re-tune your prompts — the instruction following is tighter.

If you’re building AI agents: The vision upgrade (3x resolution) and better memory across sessions make computer-use agents significantly more viable. Worth testing.

If you’re watching the AI safety space: The Project Glasswing / capability shaping approach is a genuinely new move. Agree or disagree with it, but pay attention — this is likely where the industry is heading.

🔗 Anthropic’s official announcement | Project Glasswing | Cyber Verification Program

💬 Hit reply and tell me: are you switching to 4.7 immediately or waiting? I read every response.

Until next time,

John

Here’s the thing nobody’s talking about enough: we’re not just using AI anymore – we’re building with it.

A few months ago, making your own AI agent required Python scripts, API keys, and a tolerance for broken things. Now? You can spin one up in plain English inside tools you’re already paying for.

I’ve been deep in two platforms this month that are making custom AI agents feel... normal. Here’s what I found.

Notion Custom Agents

What it is: Notion launched Custom Agents in February 2026 – autonomous AI teammates that live inside your workspace and respond to triggers (Slack messages, database changes, schedules, emails). Over 21,000 agents have already been created during the beta.

What makes it interesting:

• You describe what you want in plain English – no code

• Agents can pull from your existing Notion pages, databases, and docs as their knowledge base

• They integrate with Slack, Mail, Calendar, Figma, and more

• Scoped permissions mean you control exactly what each agent can access

Pricing: Free during beta (through ~April 2026) on Business and Enterprise plans. After that, it moves to a credits system — $10 per 1,000 Notion Credits.

My take: I built a few agents inside my own workspace – one that monitors my content calendar and drafts newsletter outlines when I add a new topic, and another that answers questions about my project docs. Setup was shockingly fast. You describe what you want, point it at the right databases, and it just... works. The trigger system is where it gets really powerful – I have one that fires whenever a new entry hits my ideas database. What surprised me most is how well it pulls context from existing pages. It’s not just keyword matching; it actually understands the relationships between my docs.

Verdict: If you’re already a Notion power user, this is a no-brainer to try while it’s free. The learning curve is basically zero.

🔗 Try Notion Custom Agents

Hyperagent (by Airtable)

What it is: Hyperagent is Airtable’s dedicated autonomous agent platform – a standalone product built to run complex, hours-long workflows across enterprise data. It gives agents a full computing environment – they can learn new tools and APIs on the fly, chain them together, and produce finished deliverables. Think: hours of research or data work, condensed into minutes.

What makes it interesting:

• Agents run autonomously with access to a real computing environment

• They learn and adapt to new tools and APIs as they go – no pre-built integrations required

• Can chain multiple tools together to complete complex, multi-step workflows

My take: What gets me about Hyperagent is the combination of breadth and depth. The integration list is genuinely impressive – it connects to the tools you’d actually want an agent touching. But the real unlock is chaining. You can wire together a sequence of complex actions and just... let it run. I’ve set it loose on tasks I’d normally block out an hour for and come back to finished work. That’s a different feeling than most automation tools, which still need you hovering. The learning curve is real – this isn’t a no-code play. But if you’re dev or dev-adjacent, that tradeoff is worth it fast.

Verdict: This is the power-user pick. If you need agents that can handle complex, multi-step work across the open web, Hyperagent is worth exploring.

🔗 Check out Hyperagent

What This Means

Both of these tools point in the same direction: AI agents are becoming personal infrastructure.

Notion is betting you’ll build agents inside your existing workspace. Hyperagent is betting you need agents that roam the open web.

Different approaches, same destination – AI that works for you in the background, not just with you in a chat window.

My prediction: by the end of 2026, everyone reading this will have at least one custom agent running. The tools are that accessible now.

Honestly, setting these up scratched the same itch as building automations in Zapier years ago – except now the automations can actually think. They read context, make judgment calls, and adapt. That’s the shift. We went from “if this, then that” to “figure out what to do and do it.” And we’re still in the early innings.

📺 Want to see these in action? I’m doing deeper walkthroughs on the YouTube channel – subscribe to Stack Snacks on YouTube so you don’t miss them.

💬 Hit reply and tell me: have you built any custom agents yet? What are you automating? I read every response.

Until next Tuesday,

John

In this post, I’m walking through three real tools I built from scratch on the Zo platform, each using a different open source model and a single prompt. Practical builds that show you what these models can actually do.

Why I’m Using Zo for This

These open source models – MiniMax, Kimi, GLM – don’t have the same plug-and-play API access as the frontier models. You need somewhere to run them, and that’s where Zo comes in.

[Zo](https://zo.computer) is a single workspace for hosting web apps, running AI models, and building tools. What drew me to it wasn’t just the model hosting – it’s that you can go from prompt to deployed app without stitching together a separate frontend, backend, and hosting stack. Everything lives in one place.

Right now, they’re offering free access to all three models, which made it easy to run real builds without worrying about burning credits. I’m on the basic paid plan, but the free tier gives you enough runway to figure out what these models can actually do.

Build #1: Link-in-Bio Page (MiniMax M2.5)

For the first test, I wanted something simple – a link-in-bio page hosted directly on my Zo space. I gave MiniMax M2.5 a single prompt with all my links (Stack Snacks, newsletter, YouTube, etc.) and told it to build the page.

**How it went:** There was a little bit of funkiness getting it to render initially. The first attempt failed, and I had to restart the chat. But once it got going, it built a clean, colorful link-in-bio page that absolutely gets the job done. It’s not going to win any design awards, but for a one-prompt build? Solid.

I also ran the same prompt through GLM-5 and Kimi K2.5. GLM-5 noticed the existing page and basically cloned it, while Kimi K2.5 was so fast it had the page up before I could even check the preview. The speed on all three models was impressive, even if the link-in-bio use case was too simple to really differentiate them.

**Verdict:** All three models handled this basic task without breaking a sweat. The real differences showed up in the next builds.

Build #2: Credit Card Rewards Calculator (Kimi K2.5)

This is where things got interesting. I wanted to see if Kimi K2.5 could do something more complex: research a brand-new credit card (the Robinhood Platinum), pull all the benefits and perks, and build me an interactive calculator to figure out if the annual fee is actually worth it.

One prompt. That’s it.

**How it went:** Kimi K2.5 went to work immediately. It ran a bunch of web searches, read through multiple articles about the Robinhood Platinum card, compiled all the offers into a markdown file, and then started building the calculator. The whole process took just a few minutes.

The result was a fully interactive tool where I could plug in my estimated spending across categories – travel credits, hotel credits, DoorDash, dining, flights, car rentals, everything else – and see whether the card’s annual fee pencils out based on my actual habits.

After running my numbers, the calculator showed me a net annual value and even compared it against the Amex Gold card. It’s exactly the kind of quick analysis I’d normally spend 30 minutes doing manually in a spreadsheet.

**Verdict:** Kimi K2.5 impressed me the most here. It handled the web research, math, and UI building in one shot. Fast, accurate, and genuinely useful output.

Build #3: AI App Tracker Database (GLM-5)

For the final build, I wanted to test GLM-5 on something with more of a coding challenge – a database-driven app. The prompt was simple: build me a tracker for all the AI apps I’m using, what I’m paying for, and a list of things I want to try. I asked for a table with tags and told it to pre-populate some of the major AI tools.

**How it went:** GLM-5 took a different approach than the other models. It built a proper little API to store the data, then created a front-end page to consume it. It even added authentication, which I hadn’t asked for – a nice touch for anything that might eventually hold real data.

The final result was a clean table interface with tags like “want to try,” “paid for,” and “stopped.” Each entry could be edited inline through an edit panel at the top of the page. You could delete entries, update statuses, and open the actual tools directly from the tracker.

Was it perfect? No – the edit functionality needed a small nudge. But for a single-prompt build that included a backend API, a JSON data store, and a responsive front-end? That’s remarkable.

**Verdict:** GLM-5 is the coding powerhouse of the group. It handled the most technically complex task without breaking a sweat.

My Final Rankings

After a few weeks of using all three models regularly, here’s where I’ve landed:

**Kimi K2.5** is my pick for general-purpose work. It’s fast, handles research tasks well, and produces clean output across a range of use cases. This one has become part of my daily workflow.

**GLM-5** is the one I reach for when I need something more technically complex. If the build involves APIs, databases, or heavier code, GLM-5 handles it with confidence.

**MiniMax M2.5** is solid across the board, but didn’t stand out as strongly as the other two in my testing. Still absolutely worth having in the rotation.

The Bigger Takeaway

Here’s what really changed for me: these models shifted *how* I allocate my AI usage. I’m now using open source models for the everyday stuff – quick tools, simple builds, research tasks – and saving my frontier model tokens (Claude, GPT-4, etc.) for the complex work that truly needs them.

If you’re a solopreneur, creator, or small business owner, this is huge. You can build real, functional tools without burning through expensive API credits or paid subscriptions. The quality gap between open source and frontier models is shrinking fast, and for a lot of daily tasks, it’s already negligible.

Try It Yourself

Head over to [Zo](https://zo.computer) and sign up for free access while it’s still available. Pick a small tool you’ve been meaning to build – a calculator, a tracker, a simple landing page – and throw a single prompt at one of these models. I think you’ll be surprised at what comes back.

If you want more AI tool breakdowns like this, check out the [Stack Snacks newsletter](https://www.stack-snacks.com/) or subscribe to the channel on YouTube.

---

*Which of these three models are you most interested in trying? I’d love to hear what you’d build with them.*

If you’ve been following the AI agent space, you’ve probably seen the buzz around running your own AI computer—setting up Mac Minis, virtual private servers, and dealing with messy configurations. It’s powerful stuff, but it’s not for everyone.

That’s where Zo comes in. It’s a tinkerer’s dream that gives you a lot of that power right out of the box. No s…

Read more

That changed this week.

Claude Cowork is here

Read more

Read more

Read more

It’s the most important shift in the AI landscape this year.

Gemini 3 Pro tops the LMSYS Arena. Claude Opus 4.5 dominates coding tasks. GPT-5.2 wins on versatility. Grok 4.1 undercuts everyone on price. Each claim is true. None of them helps you decide.

Here’s why that’s act…

Read more

ChatGPT Apps

You can now run apps directly inside ChatGPT. This means you can call up specific applications during your conversations. For example, you can bring up Canva while …

Read more

Notion just dropped version 3.0, and the headline feature is customized AI agents. The full agent product isn’t live yet, but the improvements to Notion AI are already worth paying attention to.

Here’s what’s new and what it means for how you work.

Custom Instructions: Train Your AI

The standout feature is custom instructions. You write detailed instructio…

Read more

The Superlist app's AI features for meeting notes and task management offer an intriguing solution for users looking to enhance their productivity through technology. After spending some time exploring these capabilities, it's evident that the app shines in certain areas while needing improvement in others.

Meeting Notes

One of the standout features of th…

Read more

GPT‑5 Codex is a GPT‑5 variant optimized for agentic coding workflows. It is positioned as a code-first assistant and (at least for now) isn’t available as part of ChatGPT Codex.

• What it is: A GPT‑5 build tuned for autonomous code reading, planning, and multi‑file edits

• Why it matters: Faster refacto…

Read more

“Hey there, Happy Tuesday! 👋 In this snack-sized issue, we are going over the GPT 5 release and some of our earlier thoughts.

GPT-5 Is Out

The day we've all been waiting for is here. I can't remember the last time I was so excited about a product launch that I added it to my calendar to watch the livestream. The consensus is that GPT-5 operates different…

Read more