MD at @type_a_media, a London #SEO agency famous for working a #4dayworkweek.

London, England
Opus5.5 and Jev in the last week……this is what they mean by hockey stick acceleration. Wow. 📈
101

ALT ship mar GIF by Maria Jose Guzman

Today we released the September 2026 spam update. We'll update our ranking release history page when the rollout is complete: status.search.google.com/inc…
2
203
Just changed my CI to @UbicloudHQ and upgraded to Github enterprise to get the merge queue feature. Noticeable speed increase and cost decrease. Amazing.
1
4
443
I did not foresee that Github Action minutes would ever be a development bottleneck. 😩
1
1
199
Arguably a bigger deal than the new model release 😂
Replying to @claudeai
One more thing: we’re increasing five-hour usage limits on Pro, Max, and Team plans. We’re also providing subscription users a rate limit reset, which you can save and use whenever you choose.
1
196
Trialing @opencode GO using Deepseek Flash4.1 and I am shocked at how good it is and how cheap it is. I'm getting all my daily work done and overnight sessions for about $2 a day compare to about $150 with frontiers. Adding Jev into the mix to buffer some of the decision making away from the LLMs and I honestly think its going to cost more to buy the coffee to do the vibe code session than the tokens. Brave new world
1
3
160
Deepseek Flash 4.1 is bonkers! Almost a million tokens for a cup of coffee. High quality output too. I don't see how frontier labs compete with this.
1
184
Its highly likely the Jev style 'system one' models are going to be the next big frontier in SEO and AEO. On a mid sized keyword clustering and mapping process the cost went from $96 to $0.48 💀 This will apply to other audits link kill, keep, combine and internal linking in the same way. Once its built into your continuous monitoring and crawling you're going to be a force of nature in the SERPs.
2
2
5
386
But is the quality any good? It agreed 87% of the time. And where it disagreed Sonnet also flagged as 'borderline'. Sonnet struggled when locations were introduced 'solicitors in glasgow' marked as mixed intent but Jev marked as transactional.
66
Ross Tavendale retweeted
1
3
41
ASMR goodness. Amazing copywriter available for hire! She’s even called Pen! What a name for a writer.
Got some AI drafts that could do with a juj up? I’m here to help! DM me x
1
1
230
Ross Tavendale retweeted
Got some AI drafts that could do with a juj up? I’m here to help! DM me x
1
285
Sotted in my AI app 😬
1
1
158
context: we have an audit called 'kill, keep, combine'. It's not skynet 😆
1
50
I’m fascinated to know are people running multiple accounts just shipping more features or is there a vibe coder n00b issue? For Dave doing 450 PRs I assume they are smaller PRs fixing and optimising thing and making smaller manageable features on an established code base?
What am I doing wrong? I wrote an entire app and submitted 450 PRs on a single $200 MAX plan in a month. What are people doing with thousands of dollars of tokens?
2
1
411
For me, most of my token burn is cache reads due to lots of subagent driven work on new features
2
108
Them: wahhh Fable burns too many credits, what a ripoff 😭 Also Them: 9 concurrent subagents running over 44% ctx 🙃 I am them.
1
144
Praise be!
We've just reset weekly limits for everyone on a Claude Max plan. With Fable 5.1 out and a long weekend ahead for many of you, we wanted to keep you building. I'd love to see what you build!
1
1
368
Not knowing where your AirPods are but they are already in your ears. Boomer equivalent is glasses on the head.
2
3
271
If you're running multiple AI accounts I recommend using Figma and their MCP to build out all your screens. Single source of truth and doesn't get locked behind a single login.
1
3
223