I've spent now - what? - 15 years around ads? a long time.. Longer than I expected.. esp. since I didn't target it specifically and I didn't realize it was an ecosystem at start.
My first adserver was open source and build in PHP in the early 2000s.
Those were the days...
Claude can run ChatGPT ads.
Well, mostly read them, for now. But that’s where this is heading.
A couple of weeks ago ChatGPT took their ads platform out of beta and opened it to the public. With it, they opened up not just the ads manager but their advertiser API too.
We’re somewhat of a Claude shop at https://t.co/5JFeKgfVn6, so we couldn’t help but think: can we run them in Claude? The answer was: not directly.
So we used Claude Code to build an MCP Server and bridge that gap. The model context protocol is open source and supported by both Anthropic and OpenAI.
PS: I was having a bit of a deja-vu writing this — a couple of months ago I was joking about the Claude Ads Agent in Claude Code.. and now… here we are :)
Couple of thoughts here:
1. The boring stuff takes more energy and provides less energy back / satisfaction (for most peple?!) vs the higher level tasks
2. Solving them already puts you in touch with the customer / user and start understading their pains
3. These things are less open-ended than higher level task - meaning the might be (in theory) easier to solve/verify
I would not discount the article based on just that quote alone.
Those are indeed inputs from my POV -> outputs should be:
- increase in features delivered (per week/day)
- increase in customer usage (because of said features)
- increase in customer satisfaction (lagging indicator)
HYPD is live on Product Hunt today.
The thesis: just like programming is "solved" but engineering is not, ad-ops and media buying will be solved, but account strategy and human creativity remain the leverage.
This is one framing; but the framing I prefer is around value creation;
One engineer with AI = 3x value creation than before (just an example) - for existing customers:
One sales ops = 5x value creation for client facing roles.
What is the limiting factor for revenue generation? If it’s humans, or the GDP - you can optimize everything underneath but you can’t change reality, at least not that fast (some CEOs can well distort it though). And you depend on other “cogs” of the economy.
Tl:dr efficiencies are measured against the harsh reality of the P&L which is basically the “laws of physics” for a business.
@taylorotwell Agree! Hence we laid off very few software engineers. Very few customer-facing sales people either. But if you don’t think AI changes how SalesOps, for instance, will be done going forward then it probably means you don’t know what SalesOps is.
Live from Code with Claude: we're launching dreaming in Claude Managed Agents as a research preview.
Outcomes, multiagent orchestration, and webhooks are now in public beta.
The whole podcast is really good but the end has basically my takeaways:
- things that agents can't do is your job now
- things that agents can do, they can probably do better than you
- so be strategic about what you're actually spending time on
and what I am trying to achieve myself (workwise) currently across different agents (WIP)
https://t.co/lhlxWFssYT
This is a tough post to write.
Yesterday, Claude Code's entire source code was accidentally leaked. Like everyone in tech, we spent the evening digging through it.
What we found makes HYPD obsolete.
Buried in the codebase, there's an internal module called anthropic-ads-agent. It handles campaign analysis across Google and Meta. Automated performance scoring. Budget reallocation recommendations. Even creative fatigue detection.
That's basically our current product plus most of our roadmap.
It gets worse. We found:
→ A function called generatePmaxInsights() that mirrors our Routines feature almost 1:1
→ A config file with the comment // TODO: better than HYPD
→ An internal Slack integration that sends weekly ad reports — with the exact same KPI structure we use
→ A hardcoded test campaign named "Ionut's Fake DTC Brand" with a 0.2 ROAS
After a long call between myself, Philip and Simon last night, we've made the decision: we're open-sourcing everything.
If Anthropic is building it, there's no point fighting it.
And if they are looking for a faster way to expand their Ads Team, they can reach out to [email protected].
This is a tough post to write.
Yesterday, Claude Code's entire source code was accidentally leaked. Like everyone in tech, we spent the evening digging through it.
What we found makes HYPD obsolete.
Buried in the codebase, there's an internal module called anthropic-ads-agent. It handles campaign analysis across Google and Meta. Automated performance scoring. Budget reallocation recommendations. Even creative fatigue detection.
That's basically our current product plus most of our roadmap.
It gets worse. We found:
→ A function called generatePmaxInsights() that mirrors our Routines feature almost 1:1
→ A config file with the comment // TODO: better than HYPD
→ An internal Slack integration that sends weekly ad reports — with the exact same KPI structure we use
→ A hardcoded test campaign named "Ionut's Fake DTC Brand" with a 0.2 ROAS
After a long call between myself, Philip and Simon last night, we've made the decision: we're open-sourcing everything.
If Anthropic is building it, there's no point fighting it.
And if they are looking for a faster way to expand their Ads Team, they can reach out to [email protected].
This is a tough post to write.
Yesterday, Claude Code's entire source code was accidentally leaked. Like everyone in tech, we spent the evening digging through it.
What we found makes HYPD obsolete.
Buried in the codebase, there's an internal module called anthropic-ads-agent. It handles campaign analysis across Google and Meta. Automated performance scoring. Budget reallocation recommendations. Even creative fatigue detection.
That's basically our current product plus most of our roadmap.
It gets worse. We found:
→ A function called generatePmaxInsights() that mirrors our Routines feature almost 1:1
→ A config file with the comment // TODO: better than HYPD
→ An internal Slack integration that sends weekly ad reports — with the exact same KPI structure we use
→ A hardcoded test campaign named "Ionut's Fake DTC Brand" with a 0.2 ROAS
After a long call between myself, Philip and Simon last night, we've made the decision: we're open-sourcing everything.
If Anthropic is building it, there's no point fighting it.
And if they are looking for a faster way to expand their Ads Team, they can reach out to [email protected].
@via_marketing_@amandiver@branchmetrics Yeah… 💯 - we re a small team and focus is everything so we take it channel by channel. Last week(s) have been mostly focused on outbound..
Missed this podcast going out because I was so deep in GTM last week. 🎙️🚀
The irony is not lost on me, given this is a growth podcast. 🙂 Better late than never, I guess.
Had lots of fun on this one with Adam and @amandiver from @branchmetrics. Felt like a casual conversation in a pub or at one of the usual adtech events.
We got into mobile, adtech people, competing with Facebook on native ads, baking pains (🥨), GenAI x Ads (@hypdai), when you don’t need GenAI, and even LLM Ads in dev environments. 💻
Link below 👇
Great example of where digital work is going.. And where much of the work in digital marketing is going too.
Conversational AI/UI is so much more natural for many of the daily tasks. Plus, the Google Ads API is pretty well structured and documented and plays well with agents.
Lastly, Cowork is a great unlock for tasks that don't require the full breadth and depth of Claude Code.
The question is how distribution will shape up between:
- native tooling (eg Google Ads Manager and their Advisor)
- horizontal tooling (eg Cowork)
- vertical agents (eg https://t.co/5JFeKgfVn6)?
1) my plugin ships with a few skills, including mining for search terms, budget optimizations, weekly reviews, etc. each one is a structured prompt with specific steps, evaluation criteria, and output formats based on my personal working style.