Dev Day Recap: Which AI Tools Are Dead by Q2?

I spent Dev Day watching OpenAI systematically eliminate the need for half the AI Tools I see students using.

It was phenomenal. Brutal. And exactly why I recommend folks use the native AI platform and not 12 independent tools. 

Tired of the inbox? Watch the full article on our Youtube channel, here.

Remember Plugins? The GPT Store? Yeah, everyone forgot those too.

This is different.

OpenAI is using MCP (Model Context Protocol) to render native-feeling UIs inside ChatGPT that sync directly with your tools, including Figma, Notion, Zillow, and your CRM, among others.

Why this matters for marketers:

Think about it: Do you really need to log into 12 different marketing tools? Or do you just need ChatGPT to:

All without leaving their chat.

It's the same shift we saw with mobile apps…the web didn't die, but mobile became 50%+ of most brands' traffic.

Quick Win: Audit your marketing stack this week. Which tools are going to die? Transition off them NOW. 

Tools I’m keeping: Canva, Asana, Gmail.

Tools on the bubble: Notion? I love it, but does it survive? I certainly don’t need to pay more for Notion’s AI feature. I want Notion data in my ChatGPT/Claude chats, but after conducting extensive testing with Notion MCP servers, it's currently just not working well. 

Then they dropped AgentKit, a suite of tools to add ChatGPT to your apps.

Agent Builder: A visual workflow builder (think Zapier/n8n) that lets non-technical PMs build chatbots and hand off a workflow_id to devs.

Widget Builder: Pre-built chat UI components so you don't have to design from scratch.

Translation: OpenAI wants every product to have a chat interface. And they're making it dead simple.

The uncomfortable truth: They're not just enabling startups…they're swallowing the entire "AI middleware and tool" market. Every startup building another AI tool just got a serious competitor with infinite resources.

We’re nearly ready to unveil the AI Action Club.

Marketing Takeaway: Video ads just got cheaper to produce. AI coding assistants just became table stakes for all dev teams. Voice agents just became affordable for mid-market brands.

1. Stop building dashboards, start building integrations

If your product relies on users logging in to look at data, you're on borrowed time. Start thinking: "How does this report work inside ChatGPT?"

2. The "chat-first" experience is coming for everything

Whether you're B2B SaaS or D2C ecommerce, expect your customers to want chat-based interactions. AgentKit makes it easier to build, but also easier for competitors to copy.

3. AI infrastructure is commoditizing fast

Sora, Codex, GPT-5 are all available via API. The moat isn't the models anymore. It's:

Quick Win: If you're not already collecting first-party data and documenting your best processes, start now. That's your defensible advantage.

OpenAI keeps saying: "The models are taking us wherever we're going."

Let's be real, they're not just following the models. They're training them to get better at specific things (like writing, code, building workflows) and then building products around those improvements.

For marketers: This is good news. You're about to get better tools, cheaper and faster. But as I was talking to one senior leader, "If your data is a yard sale over 15 tools, you're dead, or at the very least, you're going to need to hire an army of consultants to get you out of the mess."

Pick one marketing workflow you do manually and ask:

"Could ChatGPT + MCP + my existing tools automate this?"

Then test it.

Questions? The AI Action Club will be rolling out in November. This will be the place where we can help you go from toolageddon to MCP mastery. 

Cheers,
Alec

P.S. If you found this useful, forward it to your CTO. They need to see where this is going. If they don’t have an alpha test going with an MCP server to your LLM of choice, they are behind.