Best ChatGPT Plugins and GPTs by Industry (Q4 2026 Update)

Yes, ChatGPT plugins still exist in 2026, and the word is official again. On 9 July 2026 OpenAI replaced the App Directory with the Plugin Directory, which now bundles apps, skills and app templates into one installable package. The 2023 plugin store is not coming back, but everything it did is now handled by a system that is considerably more capable.

At the same time, custom GPTs are on their way out. OpenAI has announced it plans to retire them and move the workflows into plugins. Existing GPTs keep working until then, so this guide covers both: the ChatGPT plugins worth installing today, and the legacy GPTs that still do a job while they last.

Below is our guide to the best ChatGPT plugins and GPTs across business departments. Each category lists what to use now, what the legacy tools were, and where the practical benefit sits.

• Best ChatGPT plugins for General use
• Best ChatGPT plugins for Marketing
• Best ChatGPT plugins for Sales
• Best ChatGPT plugins for Development
• Best ChatGPT plugins for Finance
• Best ChatGPT plugins for Human Resources
• Best ChatGPT plugins for Customer Support
• Best ChatGPT plugins for Business Strategy
• Best ChatGPT plugins for Construction

Whether you want to save time on data analysis, generate marketing content, streamline sales outreach or plan a construction project, there is probably a ChatGPT plugin or GPT that fits. Let’s get into it.

What changed in 2026: the ChatGPT Plugin Directory

If you last looked at this in 2023 or 2024, the vocabulary has moved three times. Here is the short version.

Period	What it was called	Status today
March 2023 to April 2024	ChatGPT plugins (the original beta and plugin store)	Shut down on 9 April 2024
November 2023 onward	GPTs and the GPT Store	Still usable, retirement announced
October 2025 onward	Apps in ChatGPT, built with the Apps SDK on MCP	Active, now packaged inside plugins
December 2025	Connectors renamed to apps	Active
9 July 2026	Plugin Directory replaces the App Directory	Current system

The current setup is easier to explain than the history suggests. A plugin is a package. It can contain three things:

• Apps. The integration that connects ChatGPT to an outside service such as Gmail, Google Drive, Slack, HubSpot or Canva, so it can read data or perform actions.
• Skills. Reusable instructions and workflow guidance. A skill can carry examples, resources and code. ChatGPT picks up the relevant skill automatically once it is installed. Skills are available to Business, Enterprise, Healthcare and Edu accounts.
• App templates. Configuration that administrators use to set up an app for a whole workspace.

Some plugins contain several apps. Others contain only skills and never connect to an external service at all. Installing a plugin does not bypass authorisation: you still sign in to the underlying service, and workspace permissions still apply.

The practical upshot: the old plugin store was a catalogue of bolt-on tools. The Plugin Directory is a permissioned integration layer built on the Model Context Protocol, the same open standard other AI assistants use. A plugin built for ChatGPT is not locked to ChatGPT.

Are ChatGPT plugins still available in 2026?

Yes. Open chatgpt.com/plugins and you will find the current Plugin Directory, organised by category: Popular, New and Noteworthy, Small Business, Productivity, Creativity, Developer Tools, Business and Operations, Data and Analytics, Communication, Education and Research, Scientific Research, Security, Finance, Healthcare, Travel and Entertainment.

What is gone is the 2023 plugin store, the one you enabled under Settings and Beta Features and where you could run three plugins per conversation. That closed in April 2024 and its extensions were never migrated. The name returned in July 2026 for a different and better system.

GPTs vs ChatGPT plugins: what is the difference?

A GPT is a preconfigured version of ChatGPT: instructions, knowledge files and a set of capabilities, all wrapped in a persona you open from Explore GPTs. A plugin is a package of capabilities you install once, after which ChatGPT uses it inside any conversation. You do not have to leave your chat and go somewhere else.

That difference is why OpenAI is consolidating. A GPT is a destination. A plugin is a capability.

Custom GPTs are being retired

This matters if your team has built its own. The current position from OpenAI:

• Personal accounts (Free, Go, Plus and Pro) can no longer create or publish new GPTs. That changed on 16 August 2026. Existing GPTs remain usable and editable.
• New GPT creation and GPT Store publishing now require a Business, Enterprise or Edu workspace.
• OpenAI announced on 11 September 2026 that it plans to retire custom GPTs and offer a migration path to plugins.
• For affected Enterprise workspaces, retirement is planned for 11 December 2026. The migration experience was targeted for 17 September 2026. Other plans are expected to follow the same timeline.
• The official DALL·E GPT was retired on 30 August 2026. Image generation moved into ChatGPT Images.

If you rely on a custom GPT for real work, now is the time to write down what it actually does, which instructions matter, and which files it leans on. That is the raw material for a plugin or a skill.

General use (all-purpose productivity)

For general productivity the best ChatGPT tools are the ones that extend what the model can reach: your files, your calendar, your notes and your data.

Plugins to use now

• Google Drive, Dropbox and Notion. These are the workhorses. Connecting Drive means ChatGPT can work with your Docs, Sheets and Slides directly instead of waiting for uploads. Notion covers docs and workflows. Dropbox handles find, create and act. Use case: ask for a summary of last quarter’s board pack without hunting for the file first.
• Gmail and Outlook Email. Inbox triage that survives contact with a real inbox. ChatGPT reads the thread, works out what is actually being asked, and drafts a reply. Use case: clear a 40-message backlog by asking for the five that need a decision today.
• Google Calendar and Outlook Calendar. Scheduling that knows what your week already looks like.
• Granola and Fathom. Meeting notes and context. Granola adds your meeting context to a conversation, which is the difference between “write a follow-up email” and “write a follow-up email that references what was agreed”.
• Advanced Data Analysis (built in). Not a plugin at all, and still the single most useful capability for anyone working with numbers. Upload a CSV, an Excel file or a messy PDF and ChatGPT runs Python in a sandbox to analyse it and plot the result. Use case: “analyse this sales CSV and plot revenue by month” returns a summary and a chart without opening Excel.

Legacy GPTs and plugins (still usable for now)

• Advanced Data Analysis GPT, the original wrapper around the same built-in capability.
• Wolfram GPT, still the reference point for verified computation. It handles symbolic equations, curated scientific data and high-fidelity plots, all in natural language. Wolfram remains listed under Education and Research in the Plugin Directory.
• Productivity Tools GPT, the format-conversion middleware: scanned PDF to Excel, video to GIF, batch document processing.
• AskYourPDF GPT, for question answering over long PDF reports.
• Browser Pro, successor to the old Link Reader plugin, for fetching and summarising live URLs and YouTube transcripts.
• Consensus GPT, which connects ChatGPT to around 200 million academic papers. Consensus is now also a plugin in the directory, so this one made the jump.

Marketing

Marketing teams need content, design and SEO insight. In 2026 the useful shift is that ChatGPT can now produce the asset rather than describe it.

Plugins to use now

• Canva. The most immediately satisfying plugin in the directory. It creates, reviews and edits real Canva designs from the conversation. Use case: turn an approved blog post into a five-slide carousel with your brand template applied, then change the copy on all five without opening Canva.
• Figma. Creates designs and ships to code. Where Canva produces something close to finished, Figma produces a file a designer will keep working in. Good for flowcharts, wireframes and structure.
• Ubersuggest. Keyword research and SEO insight inside the chat. Use case: pull keyword volumes for a content brief and draft the outline in the same conversation.
• Metricool and Windsor.ai. Social scheduling and cross-channel campaign data. Windsor.ai connects over 350 data sources, which makes it the practical answer to “why did last week’s numbers move”.
• vidIQ. YouTube stats and keywords, for anyone running a channel.
• Higgsfield, Runway, invideo and OpenArt. Image and video generation from inside ChatGPT.

Worth saying plainly: the SEO game itself has changed. Ranking in Google is no longer the only target, because a growing share of research now happens inside AI assistants. If you want to be cited when someone asks ChatGPT about your category, the work is structural: clear claims, clean headings, specific numbers, and content that answers a question rather than circles it.

Legacy GPTs (still usable for now)

• SEO Core AI, which audits a URL for both traditional SEO and AI visibility.
• E-commerce Listing GPT, for Amazon, Etsy and Shopify product descriptions built around high-traffic keywords.
• Copywriter GPT, for ad copy, social hooks, email sequences and landing page copy in a consistent brand voice.
• PostEverywhere GPT, which drafts and schedules across platforms.
• GPTHuman AI, a humaniser that adjusts rhythm and tone so marketing copy stops sounding like a machine wrote it.

Sales

For sales teams, the value is in removing the admin around selling: research, enrichment, CRM hygiene and follow-up.

Plugins to use now

• HubSpot. Insights to action, directly in HubSpot. A CRM answers “what is this deal worth” easily. It answers “what do our last twelve losses have in common” slowly, if at all. That second question is the one worth connecting for.
• Lusha and ZoomInfo. Verified B2B contact data and buying signals, for building a list that is actually reachable.
• Shopify and Stripe. Store management and payments. For commerce teams, asking about revenue movement and getting an answer from live data beats waiting for a weekly report.
• Slack and Teams. Read, summarise and act on internal threads. Use case: find every mention of an account across channels before a renewal call.
• Superhuman Mail. Email and calendar assistance, aimed at getting an inbox back under control.
• monday.com. Projects, tasks and CRM in one connection.

Legacy GPTs (still usable for now)

• Sales Sidekick GPT, built around deal orchestration and buyer-intent monitoring.
• Sales Relationship GPT, which scans earnings calls and leadership changes to spot high-intent windows in Tier 1 accounts.
• Adzviser, which pulled marketing and sales data from over 50 platforms into the chat for conversational reporting.
• Reclaim AI GPT, for calendar defence and focus time.
• LeadGenius, formerly LeadBuddy, for qualifying leads against an ICP before outreach.

Development (engineering and IT)

Development is where the 2026 plugin system pulls furthest ahead of the old plugin store, because the integrations now write as well as read.

Plugins to use now

• GitHub. Triage pull requests, issues and CI, and handle publish flows. Use case: “summarise what changed in this release branch and flag anything that touches auth”.
• Supabase and Firebase. Manage and query databases, build and manage apps.
• Vercel and Railway. Build and deploy web apps and agents.
• Firecrawl and Exa. Web search and scraping built for agents, including document parsing and YouTube transcripts. These have largely replaced the old browsing plugins.
• Codex Security. Security scanning for your codebase.
• Advanced Data Analysis (built in). Formerly Code Interpreter. Still the fastest route to writing, executing and self-debugging Python without setting up a local environment. Use case: upload raw files and have the join logic and schema validation handled in one pass.

Legacy GPTs (still usable for now)

• Grimoire, the prompt-programming GPT for building sites and apps from natural language, with vision-to-code from a sketch or screenshot.
• Python, Java and JS Expert GPTs, with slash commands like /fix and /tests for debugging and unit tests.
• Documentation GPTs such as API Docs, which fetch live documentation instead of relying on stale training data.
• Eraser (DiagramGPT), successor to Show Me, for diagram-as-code architecture diagrams you can version control.

Finance

Finance has moved from calculators to connected analysis. This is also the category where accuracy matters most, so favour plugins that read from a real source over ones that generate plausible numbers.

Plugins to use now

• Intuit QuickBooks. Business finances, in conversation. For an SME this is the highest-value connection in the whole directory.
• Interactive Brokers (IBKR). Analyse global markets from account data.
• Alpaca. Market data for stocks and crypto.
• Public Equity Investing. Public equity research.
• Parqet. Portfolio analysis, popular in the Dutch and German market.
• Stripe. Revenue and payments data.
• Tableau, Microsoft Power BI, ClickHouse and AWS Data Analytics. For finance teams sitting on a warehouse rather than a spreadsheet. Note that the Tableau MCP-only listing is marked for deprecation, so check which one you are installing.

Legacy GPTs (still usable for now)

• Boolio Invest, the stock analysis GPT that returned quantitative factor breakdowns from a ticker.
• Macro Economics Expert, for central bank and trade data analysis.
• Finance and Economics GPT, covering stocks, crypto and macro indicators.
• AskYourPDF (Financial Edition), for parsing 10-Ks and dense filings.
• Magnifi GPT, a conversational investment assistant that replaced PortfolioPilot.
• Wolfram GPT, still the gold standard for financial mathematics that has to be correct rather than merely confident.

Human Resources

HR plugins save time on screening, onboarding and policy work. Two cautions worth stating: candidate data is personal data, and automated screening has legal limits in the EU. Connect deliberately, and keep a human in the loop on decisions.

Plugins to use now

• Gusto. Payroll and team insights.
• Google Drive or Notion. Where your handbook, policies and templates already live. This is usually the highest-value HR connection, because it turns the policy set into something answerable.
• Kahoot! and Quizlet. Training material, instantly. Use case: turn a safety manual into a set of quizzes new hires actually complete.
• AI Roleplay Chat Simulator. Practice conversations and dialogue, useful for interview and performance-conversation rehearsal.
• Zoom, Fathom and Granola. Interview and one-to-one notes with structure.

Legacy GPTs (still usable for now)

• CV and Resume Evaluator, which compared resumes against a job description and scored suitability.
• Human Resources Advisor GPT, for manager and employee questions and workforce modelling.
• Employee Training GPT, which built branching training scenarios with immediate feedback.
• Final Round AI, originally candidate-side interview prep, used by HR teams to benchmark their own hiring bar.
• Enboarder GPT, for onboarding journeys.
• Employee Sentiment Analyzer, for reading survey responses at volume.

Customer Support

Support is about response quality at speed, plus learning from what customers keep telling you.

Plugins to use now

• Slack and Teams. Internal escalation, thread summaries and handover notes.
• Zoom. Meeting insights from support and escalation calls.
• Gmail and Outlook Email. Draft on-brand replies grounded in the actual thread history.
• Notion or Google Drive. Connect the knowledge base so answers come from your documentation rather than the model’s general knowledge. This one change removes most of the hallucination risk in support use.
• PostHog, Mixpanel and Amplitude. Product analytics, so you can check whether the thing a customer is complaining about is happening to everyone.

Legacy GPTs (still usable for now)

• Support Copilot GPT, which drafted policy-compliant resolutions from customer history.
• Sentiment Analysis GPT, for clustering thousands of survey responses into themes rather than scores.
• AutoGPT Agent, for autonomous multi-step support workflows.
• Fin Knowledge Base (Intercom) and Unbabel GPT, for grounded answers and multilingual support operations.

Business Strategy

For strategy work, the job is research at a depth a person would need a week for, with sources you can check.

Plugins to use now

• Consensus, SciSpace and Undermind. Research grounded in published papers rather than blog posts. Consensus covers around 200 million papers. Undermind finds and reads research. Use case: validate a market assumption with something more solid than a competitor’s landing page.
• Firecrawl. Structured competitive intelligence from sites that resist scraping.
• Tableau, Power BI, Sigma and ThoughtSpot Spotter. Bring the actual numbers into the strategy conversation.
• Canva or Figma. Turn the finished analysis into something presentable.
• Deep research (built in). Before installing anything, remember that ChatGPT’s built-in deep research already runs multi-source investigations with citations.

Legacy GPTs (still usable for now)

• Market Research and Trend Analyst, for deep-dive competitive and market analysis with SWOT output.
• AI User Persona Generator, for synthetic personas and concept testing.
• SlideSpeak GPT, which converted long strategy documents into editable PowerPoint decks.
• Firecrawl and ScrapeGraph AI, for competitor monitoring.

Construction

Construction is the clearest example of a category the Plugin Directory does not serve well yet. There are no construction-specific plugins in the official directory, which means the sector is running on built-in capabilities, legacy GPTs and custom integrations.

What works now

• Advanced Data Analysis (built in). Takeoff quantities, cost comparisons and schedule analysis from spreadsheets and exports.
• Google Drive or Dropbox. Connect the project folder so drawings, specifications and RFIs are answerable.
• Vision on drawings and photos. Upload a plan or a site photo and ask about it directly. This is a built-in capability, not a plugin, and it has quietly become the most useful feature on site.
• A custom plugin or MCP server. For anything that touches Procore, Autodesk, Revit or a local ERP, the route is a custom integration. This is where most real construction value sits.

Legacy GPTs (still usable for now)

• Construction Project Planner GPT, for multi-phase scheduling and what-if simulations on delays.
• Construction QA and Codes GPT, for jurisdiction-aware building code questions with clause citations.
• Construction Estimator GPT, for stress-testing bids and spotting missing line items in an RFP.
• HandsHQ (Safety Trainer GPT), for site-specific toolbox talks and safety briefings.
• STACK and Kreo, for automated quantity takeoffs against Revit or CAD models.
• Globality and Inventive AI, for procurement and supplier quote analysis.

How to install and use ChatGPT plugins

The process is short, which is the point.

1. Open ChatGPT on web or desktop and select Plugins in the sidebar to open the Plugin Directory.
2. Select a plugin and review what it contains: its skills, its required apps, and what setup it needs.
3. Select Install plugin. This may start the app’s authorisation flow.
4. Sign in to the underlying service when prompted. Installing a plugin never grants access the account itself does not have.
5. Use it by typing **@** in a conversation and naming it, or let ChatGPT pick it up automatically when your request matches.

A few things worth knowing before you start:

• Availability depends on your plan, workspace, role and region. The directory is visible on all plans, but installing or invoking a specific plugin is not. Business workspaces have apps enabled by default. New Enterprise and Edu workspaces start with a selected set.
• Plugins are free to connect. You pay for your ChatGPT plan and for the underlying tool’s own subscription. The connection itself costs nothing.
• Permission controls are per app. Since June 2026 you can choose when a connected app asks for permission: always, before making changes, or only before important changes.
• Some plugins are marked Desktop only and will not run in ChatGPT on the web.
• Look for the OpenAI Verified badge, but do not treat it as a security review. It signals quality and reliability, not that the plugin meets your organisation’s privacy requirements.

Building your own ChatGPT plugin

The reason the directory looks thin in some categories is straightforward: it covers horizontal software, not your sector. There is a plugin for Slack. There is no plugin for your ERP, your permit system, your claims platform or your production planning.

That gap is where most of the actual value sits, and it is now reachable without a special relationship with OpenAI. Because plugins run on the Model Context Protocol, an MCP server you build for ChatGPT also works with Claude and other MCP-aware assistants. You build the integration once.

Three routes, depending on what you need:

• A skill. No external service, just reusable instructions, examples and code for a workflow your team repeats. This is the cheapest thing to build and often the most underrated. Think of a skill as a versioned operating procedure rather than a clever prompt: define the inputs, the outputs, the permissions and a safe fallback.
• A custom connector or MCP server. For reading from and writing to your own systems. Available on paid plans through developer mode, or workspace-wide via an imported plugin marketplace.
• A workspace plugin. Package the skills and apps together, publish to your workspace directory, and set an installation policy so the right roles get it automatically.

This is the work DataNorth does for organisations: taking a process that currently runs on copy-paste and turning it into something a colleague invokes by asking. If you are weighing up whether a custom plugin, an MCP integration or an agent is the right shape for a specific process, get in touch.

Conclusion

The ChatGPT plugin ecosystem in 2026 is in better shape than the naming chaos suggests. Plugins are official vocabulary again, the directory covers most horizontal business software, and the integrations act rather than just fetch. Custom GPTs are winding down, which is a real migration task for teams that built on them, but the replacement is more useful than what it replaces.

The advice we give clients has not changed much: start by connecting the systems your work already runs on, then look at whether the remaining friction is worth a custom plugin. The biggest wins are rarely the most impressive-sounding tools. They are usually the moment ChatGPT can finally see the file, the ticket or the order without anyone pasting it in first.

Frequently asked questions (FAQ) about ChatGPT Plugins & GPTs