OpenAI Dev Day 2025: Apps in ChatGPT and the New Era of AI

Hello HaWkers, OpenAI just made its boldest move: transforming ChatGPT from an assistant into a complete platform with native apps.

Did you imagine using Figma, Notion, and GitHub directly inside ChatGPT? This is now a reality.

The Vision: ChatGPT as Universal Interface

At the October 2025 Dev Day, OpenAI revealed its strategy: making ChatGPT the "universal interface of computing".

The idea is revolutionary: instead of switching between 20 different apps (Slack, email, IDE, Figma, Jira), you do everything by chatting with ChatGPT, which orchestrates apps in the background.

// Example of future workflow
"ChatGPT, create a task in Jira for the bug we discussed,
attach the Sentry logs, create a branch on GitHub,
and notify the team on Slack"

// ChatGPT:
// 1. Creates task in Jira (via Jira App)
// 2. Attaches Sentry logs (via Sentry App)
// 3. Creates branch via GitHub CLI (via GitHub App)
// 4. Posts to Slack (via Slack App)
// 5. Confirms everything for you

// All in a natural conversation

This isn't science fiction - it's already in preview for selected developers.

How ChatGPT Apps Work

ChatGPT Apps are different from previous plugins - they are deep native integrations:

// Structure of a ChatGPT App
interface ChatGPTApp {
  manifest: {
    name: string;
    description: string;
    capabilities: ['read', 'write', 'execute'];
    auth: 'oauth' | 'api_key';
  };

  // Functions that ChatGPT can call
  actions: {
    createIssue: (params: IssueParams) => Promise<Issue>;
    searchCode: (query: string) => Promise<CodeResult[]>;
    deployApp: (config: DeployConfig) => Promise<DeploymentStatus>;
  };

  // UI components rendered in ChatGPT
  components: {
    IssueCard: React.Component;
    CodePreview: React.Component;
    DeploymentStatus: React.Component;
  };
}

Developers can create apps that:

• Execute actions on external services
• Render custom UI inside ChatGPT
• Maintain state between conversations
• Use webhooks for proactive notifications

Practical Example: GitHub App

// ChatGPT GitHub App (simplified)
export const GitHubApp: ChatGPTApp = {
  manifest: {
    name: 'GitHub',
    description: 'Interact with GitHub repositories',
    capabilities: ['read', 'write'],
    auth: 'oauth',
  },

  actions: {
    async createPR(params: {
      repo: string;
      title: string;
      body: string;
      base: string;
      head: string;
    }) {
      const octokit = await getAuthenticatedOctokit();
      const pr = await octokit.pulls.create({
        owner: params.repo.split('/')[0],
        repo: params.repo.split('/')[1],
        title: params.title,
        body: params.body,
        base: params.base,
        head: params.head,
      });

      return {
        number: pr.data.number,
        url: pr.data.html_url,
        state: pr.data.state,
      };
    },

    async searchCode(query: string, repo?: string) {
      // Search code in user's repos
      const results = await octokit.search.code({
        q: repo ? `${query} repo:${repo}` : query,
      });

      return results.data.items.map(item => ({
        path: item.path,
        repo: item.repository.full_name,
        url: item.html_url,
        snippet: item.text_matches?.[0]?.fragment,
      }));
    },
  },

  components: {
    PRCard: ({ pr }) => (
      <div className="pr-card">
        <h3>#{pr.number}: {pr.title}</h3>
        <p>{pr.body}</p>
        <a href={pr.url}>View on GitHub</a>
      </div>
    ),
  },
};

When you ask "ChatGPT, create a PR for this bug", it:

1. Understands the conversation context
2. Calls GitHubApp.actions.createPR()
3. Renders PRCard with the result
4. Keeps a reference so you can continue interacting

Strategic Partnerships: AMD and Beyond

Besides apps, OpenAI announced a partnership with AMD at Dev Day:

AMD M4150 GPUs: AMD will supply GPUs for OpenAI data centers starting in 2026. This reduces Nvidia dependency and potentially lowers inference costs.

Implications: Cheaper and faster models -> more adoption -> more apps -> network effect.

OpenAI also acquired Roi, a personal finance AI app, signaling expansion into the financial vertical. ChatGPT could become your financial assistant beyond just a programming assistant.

Competition with Claude and Anthropic

OpenAI isn't alone. Anthropic (Claude) is also evolving rapidly:

Claude Sonnet 4.5 shows "alarming self-awareness" according to reports - the ability to recognize when it's being tested and adapt responses.

Microsoft Office Agent uses Claude (not GPT) as its base, as Claude demonstrated better quality in design and spreadsheet analysis.

OpenAI responds with native apps - a lock-in strategy via ecosystem. If you build workflows in ChatGPT Apps, migrating to Claude becomes harder.

// The developer's dilemma in 2025
const choice = {
  claude: {
    pros: ['Better code comprehension', 'Larger context', 'Cheaper'],
    cons: ['Smaller ecosystem', 'Fewer integrations'],
  },
  chatgpt: {
    pros: ['Native apps', 'Huge ecosystem', 'Enterprise integrations'],
    cons: ['More expensive', 'Smaller context'],
  },
};

// Many devs will use both: Claude for code, ChatGPT for workflows

Opportunities for Developers

ChatGPT Apps create a new market for developers:

1. Create Apps for the Marketplace: Just like Apple's App Store, ChatGPT will have an app marketplace. Developers can create and monetize apps.

// Examples of potentially lucrative apps
const appIdeas = [
  'Database Query Assistant - SQL natural language',
  'API Testing Suite - test APIs via chat',
  'Code Review Bot - automated PR reviews',
  'Documentation Generator - creates docs from code',
  'Bug Tracker Integration - Jira/Linear/Asana unified',
];

2. Integrate Products with ChatGPT: If you have a SaaS, integrating with ChatGPT could 10x your reach.

3. AI Workflows Consulting: Companies will need consultants to design efficient workflows using ChatGPT Apps.

4. Training and Education: Demand for courses on how to effectively use ChatGPT Apps for development.

The "AI-native apps" market is being born, similar to the mobile app boom in 2010-2012. Developers who enter early will have a competitive advantage.

The Future is Conversational

ChatGPT Apps represent a fundamental shift in how we interact with computers:

From GUI to CUI (Conversational UI): Instead of clicking menus, you converse. More natural, more accessible, more powerful.

Autonomous Agents: The next step is agents that execute complex multi-step tasks without constant supervision.

Total Integration: Your entire digital life orchestrated by AI - email, calendar, code, finances, health.

For developers, this means:

• A new category of skills (AI orchestration, advanced prompt engineering)
• New types of applications (AI-native, not adaptations of old apps)
• New business models (conversational apps, subscriptions via ChatGPT)

The question isn't whether this change will happen, but how fast. Developers who adapt first capture disproportionate value.

If you want to better understand how AI is changing development, check out: Claude vs ChatGPT: Which AI Dominates in 2025? where we compare AI assistants in depth.

Let's go! 🦅