Updated September 27, 2026.
Hey, curious people.
I’ve been looking at beehiiv, the software I use for this newsletter. One thing about its pricing has stuck with me: as a publisher’s audience grows, their bill grows too, without changing plans. That turned into this week’s teardown.
On the menu this week:
🖥️ Anthropic says Opus 5.5 costs 40% less to run. That’s its number, not mine.
🛠️ The play: hand a Jira task to AI and get back what’s done, drafted and blocked.
🔍 Growth teardown: beehiiv’s bill grows when its customers do.
📡 On the radar: more work apps are landing inside Gemini.
01 / THE BIG MOVE
🖥️ Anthropic says Opus 5.5 costs 40% less to run
Anthropic released Claude Opus 5.5 on September 22 and put cost in the foreground. Anthropic’s release.
The number that matters: 40%. That’s how much less Anthropic says typical workloads cost than on Opus 5, at default settings. It’s a vendor number, not a saving I’ve measured.
What shipped:
Lower list prices: $4 per million input tokens and $20 per million output, 20% below Opus 5.
Fewer tokens per task, according to Anthropic. That’s where the rest of the 40% comes from.
It’s available on Anthropic’s platform and through AWS, Google Cloud and Microsoft Azure.
Why it matters for GTM: A lot of GTM AI work is recurring: research an account, enrich a list, draft the follow-up. Every run has a cost, so price per run decides which workflows stay switched on. RevOps and whoever owns the AI budget will feel this first. A workflow that was too expensive to run on every account might clear the bar now. Then the question moves from “can we afford to run it” to “does the output hold up.”
TRY THIS ON ONE TASK
Rerun one recurring task on the new model.
Same input, acceptable output. Compare the total cost and the time you spend fixing the result. A cheaper run earns its place when the result still holds up.
02 / PUT IT TO WORK
🛠️ The play: hand a Jira task to AI
I hate being in Jira. Any moment I’m spending inside the Jira interface is a moment too long for me. Its MCP, though? I really like it.
Bring back the work, along with what still needs a person.
MCP is the connection that lets an AI tool work with Jira. My morning workflow reviews my to-dos, then has AI take each one as far as it can. The full setup is in From helpful answers to finished work.
THE PLAY AT A GLANCE
Good for: anyone whose to-dos live in Jira.
You’ll need: an approved Jira connection with read and write permissions, plus connections to wherever the data and the finished work live.
Setup: Atlassian’s setup guide explains the connection.
Status: part of my morning workflow.
How it runs:
Pick one task with a clear finish line. For a reporting task, that might be a saved report linked from the Jira issue, with its sources and date range.
Ask it to bring back what it completed, what it prepared and what is blocked.
Check the actual output before marking the task done.
Copy the prompt:
Look at [Jira issue key].
Take it as far as you can.
When you stop, report back:
1. Completed, with links.
2. Prepared, waiting on me.
3. Blocked, and what you need.
Don't mark the issue done.
I'll check the output first.Keep it if checking and correcting the result takes less effort than doing the task yourself.
03 / GROWTH TEARDOWN
🔍 beehiiv: the bill grows when the customer does
On its monthly pricing page, Scale is $49 for 1,000 subscribers and $169 for 25,000. So beehiiv has a reason to help an existing customer find more readers, even if that customer never moves to a different feature plan.
Scale at 25,000 subscribers: $169 a month. At 1,000: $49. Captured September 26, 2026.
THE SHORT VERSION
How it makes money: plans priced by audience size. The same Scale plan costs more as the list grows.
Where growth comes from: existing publishers adding readers, with no plan change needed.
What I’d test: helping finance YouTubers with engaged audiences launch newsletters, with AI preparing the launch brief and a person running one focused session.
The catch: beehiiv charges for audience size, not profit, so a bigger list can mean a bigger bill without a healthier business. Still, it gives the company a concrete reason to help customers make progress.
Borrow it: Start with one obstacle that keeps your customers from getting more value. Then ask whether removing it could also improve retention or expansion. That’s a more useful starting point for a webinar than “we should probably do more webinars.”
04 / PIPELINE SIGNALS
📡 On the radar
Google began rolling out new Connected Apps in Gemini on September 23, including Airtable, Linear and monday.com, plus creative tools such as Adobe and Webflow. So what: if you already use one, look for it in Gemini’s settings and see whether its supported actions can finish a task that currently sends you between apps.
OVER TO YOU
💬 Where do your customers get stuck?
What’s one thing your customers struggle with that your team could help them get past? Hit reply. I’m interested in the obstacles that deserve more attention, especially when solving them could help both sides grow.
How was this week’s issue?
Robert
P.S. Know someone trying to put AI to work in go-to-market? Forward this to them. They can subscribe at bizgrowth.ai.