We highly recommend you watch the opening keynote here
Your Problem: Building AI tools takes too long, costs too much, and requires specialists you don’t have.
OpenAI’s Answer: They just made it faster and cheaper. Here’s what changed and why it matters.
The Numbers Tell the Story
ChatGPT went from 100 million users to 800 million in one year. That’s not hype. That’s your customers, employees, and competitors already using AI daily.
OpenAI’s developers doubled to 4 million. Their systems now process 6 billion tokens per minute, up from 300 million. (Tokens are how AI reads and writes text. More tokens means faster, bigger AI tasks.)
What this means: AI moved from experiment to standard business tool. Fast.
What OpenAI Announced
Apps Inside ChatGPT
The Problem: Your customers use ChatGPT, but they have to leave it to use your tools.
The Solution: Apps SDK (a developer toolkit) lets you build your app directly inside ChatGPT. Your users never leave the chat window.
Real Example: Coursera built their entire course catalog into ChatGPT. Students find classes, enroll, and start learning without opening another tab. Canva did the same for design. Zillow for real estate searches.
Why It Matters: ChatGPT has 800 million weekly users. You can reach them where they already are.
AgentKit for Building AI Agents
The Problem: Building AI agents (tools that complete tasks for you) is complex. Testing them is harder. Knowing if they work is guesswork.
The Solution: AgentKit gives you a visual builder, testing tools, and automatic performance tracking in one package.
Real Example: Albertson’s uses it to help store managers. The AI answers questions about inventory, scheduling, and operations. HubSpot uses it to make their customer service AI more accurate.
Why It Matters: You can build and test AI agents in days instead of months. You’ll know if they work before you launch them.
Codex Coding Agent
The Problem: Your developers spend hours on repetitive code. Projects take months.
The Solution: Codex writes code alongside your team. It’s now fully available after testing, with Slack integration.
Proof It Works: Developers using Codex complete 70% more tasks each week.
Why It Matters: Your team ships products faster. They spend time solving hard problems, not writing boilerplate code.
GPT-5 Pro and Sora 2 Available Now
The Problem: The best AI models were locked inside OpenAI’s own products. You couldn’t use them.
The Solution: GPT-5 Pro (their smartest model for complex tasks) and Sora 2 (creates videos from text) are now available through their API. Any developer can use them.
Real Example: Mattel uses Sora 2 to turn product ideas into videos in minutes. No film crew needed.
Why It Matters: You can build products with the same AI that powers ChatGPT. Your competition can too.
What This Means for Agent Companies
If you’re building agent platforms like Smythos, here’s your reality check:
OpenAI just made agent building easier and faster. That’s good and bad news.
The Threat: OpenAI has 800 million users and 4 million developers. Their AgentKit works inside the most popular AI tool on earth. They can reach customers you can’t.
Your Advantage: You can still win, but only if you solve specific problems better.
How to Compete:
Go Narrow and Deep: OpenAI builds for everyone. You can build for specific industries. A legal agent platform will beat a general one for law firms. An accounting agent platform will beat a general one for CFOs.
Make It Work Out of the Box: OpenAI gives developers tools. You can give executives solutions. Pre-built agents that solve specific problems beat blank canvases.
Control and Compliance: Many companies won’t put sensitive data in ChatGPT. Offer deployment in their own environment. Add industry-specific compliance tools. Give them control OpenAI can’t.
Integration Matters: ChatGPT connects to many tools, but not everything. Build deep integrations with the systems your customers actually use. Connect to their CRM, ERP, and data warehouses seamlessly.
Support and Training: OpenAI has documentation. You can have experts. Companies will pay for someone who understands their business and helps them succeed.
The Bottom Line: General platforms win on features. Specialized platforms win on results. Choose your niche and serve it better than anyone.
What You Should Do Next
If You’re an Executive:
- Test ChatGPT apps from Coursera, Canva, or Zillow. See how integrated AI works.
- Ask your team: “What tasks take us the most time?” Those are agent opportunities.
- Look at AgentKit. If you’re building agents, it might cut your timeline in half.
If You’re Building Agent Platforms:
- Pick one industry or use case. Own it completely.
- Study what OpenAI released. Understand what’s now commodity and what’s still differentiated.
- Talk to ten potential customers this week. Find the problems OpenAI’s tools don’t solve.
If You’re a Developer:
- Try Apps SDK. Build something small inside ChatGPT.
- Test Codex on your next project. Measure the time saved.
- Experiment with GPT-5 Pro. Learn what it does better than other models.
The Real Story
OpenAI just lowered the barrier to building AI products. That means more competition for agent platforms. But it also means more customers ready to buy.
The companies that win will solve specific problems better than anyone else. Not because they have the best AI. Because they understand their customers’ actual needs.
AI tools are commoditizing. Business understanding is not.
One More Thing: If you’re waiting for AI to stabilize before you start, you’ll wait forever. These changes happen every few months now. Start small. Learn fast. Adjust as you go.
That’s how you win in a market that changes this fast.
About Relevant Intelligence: We help executives use AI to solve actual business problems. No hype. No complexity. Just results. Want to see how these new OpenAI tools apply to your specific situation? Let’s talk.