Summary

Four short years after the first “wow” demos of GPT-3 and DALL·E, generative AI has settled into our daily products. Most software now assumes an AI layer for search, writing, analytics, or customer support. When thinking about the current landscape in 2025 we can break it into thirteen practical domains.

If you work in the AI field, you will most likely be building in one of these domains. As you kick off you're project for this course, think about which domain are you interested in building for.

Above video is from a few months ago, below is a more comprehensive 2025 product list:

1. General Assistants & Search

Large language models (LLMs) remain the Swiss-army knives of the stack.

• ChatGPT (now GPT-4o) and Gemini 2.5 Pro solve everyday questions in text, image, or voice

• Claude devours book-length inputs and used as state of the art for coding

• Grok and Perplexity answer best with citations

• Meta AI is integrated inside Facebook, Instagram and WhatsApp

• Apple Intelligence runs a pared-down model entirely on-device for privacy-first tasks.

• Almost all use-cases focus now on multimodal, real-time, and increasingly personalized.
  
  

2. Productivity & Office

If LLMs are the brain, office co-pilots are the muscle.

• Microsoft 365 Copilot drafts and audits documents and its new Copilot Actions will even book flights from inside Word.

• Duet AI brings one-click writing and image generation to Gmail and Slides

• While Notion AI for Work turns messy meeting notes into clean proposals.

• Curio builds “get-to-know-you” briefs from LinkedIn profiles;

• GrammarlyGO polishes tone

• Otter.ai records, transcribes and extracts action-items on the fly.

• Granola takes great meeting notes and making them actionable
  
  

3. Enterprise Marketing & Content

Marketing teams now expect copy at the speed of social. Enabling creative execution in minutes.

• Jasper creates campaign assets tuned to audience segments

• Writer locks output to a corporate voice guide

• Copy.ai, Anyword, Writesonic and MarketMuse balance SEO with brand tone

• Google’s new Generative Ads Studio can build YouTube ads straight from a product URL

  
  

4. Workflow Automation

“No-code” has evolved into “no-prompt beyond plain English.”

• Zapier AI Copilot suggests automations you didn’t know you needed

• UiPath Autopilot translates a paragraph of instructions into a full RPA bot

• Bardeen records a manual browser routine once, then replays it forever

• Make.com offers a visual canvas with AI triggers.
  
  

5. Creativity & Media

Generative models are eating the creator stack:

• Images: Midjourney, DALL·E 3, Adobe Firefly, and the open-source Stable Diffusion cover everything from mood boards to production assets.

• Video: Google’s Veo 3 generates scenes with synced audio; Runway, Pika Labs and Luma Dream Machine handle shorts, edits and cinematic shots. Flow: Google’s “AI co-director,” plans camera moves and storyboards.

• Audio & Voice: ElevenLabs clones voices; Suno, Udio, Mubert and Murf compose full tracks or voice-overs.

• 3-D & Design: Meshy AI outputs printable 3-D meshes

• Canva Magic, Figma AI and Gamma build brand kits, UIs or slides in seconds

• Captions: Enabling influencer video creations in seconds
  
  

6. Companions & Characters

Millions now chat daily with AI confidants.

• Replika and Inflection Pi lean toward emotional support

• Character.AI hosts role-play with any persona

• Snapchat’s “My AI” gives instant support to snap users

• Dot showcased great UX with personalized companion experience

  
  

7. Education & Research

AI tutors are moving from extra credit to core curriculum.

• Khanmigo explains algebra step-by-step

• Duolingo Max role-plays café orders in Spanish

• Google’s Notebook LM lets students chat with their own notes (now in a mobile app)

• Phind, Consensus, Semantic Scholar and Elicit reduce hours of literature review to minutes of Q & A

• DeepResearch across general assistants helps many academic and related fields

8. Developer Tools

Coding assistants now operate at the pull-request level.

• GitHub Copilot X writes tests and explains compiler errors

• Cursor answers questions about your codebase

• Replit Ghostwriter builds and deploys full stacks in-browser

• Devin promise autonomous bug-fixing and feature shipping

• Claude Code and Tabnine fill niches in secure or command-line workflows

• Hugging Face Hub hosts 1.7 million open-source models

• Lovable lets non-coders launch apps through chat.

9. Customer Support

• Zendesk Advanced AI, Intercom Fin, LivePerson, Ada and Freshworks Freddy now resolve Tier-1 tickets instantly, push empathetic follow-ups, and hand complicated cases to humans; shifting support agents to true problem-solvers.

10. Data Analysis & BI

Analytics finally speaks human.

• Tableau GPT, Looker + Gemini, Julius AI, Hex, Polymer and DataRobot let anyone ask, “Show churn by cohort,” and see a chart plus plain-English narrative. Dashboards build themselves; insights arrive conversationally.

11. Business & Enterprise AI

From CRM to compliance, AI is now baked into workflows:

• Salesforce Einstein GPT, Microsoft Viva Sales, HubSpot ChatSpot, IBM watsonx and Gong pair LLM muscle with audit trails, bias checks, and secure data pipelines.

12. Healthcare & Compliance

Regulated sectors adopt cautiously but decisively.

• DAX Copilot drafts clinician notes

• Doximity provides AI diagnostics for physicians

• Harvey scans legal precedents

• SecureAI Guardian audits ML models

• Even the FDA’s internal “Elsa” model now summarises drug-trial dossiers.

  
  

13. Agentic & Autonomous Systems

We’re entering the age of goal-seeking agents:

• OpenAI Agents SDK, AutoGPT and BabyAGI spin up constrained, tool-using helpers.

• LangGraph + LangSmith orchestrate production-grade multi-agent workflows.

• AgentForce builds autonomous Salesforce reps.

• Windows’ new Model Context Protocol ensures agents can tap native apps safely.
  
  

Trends to watch as new players enter the race

1. Multimodal everything – text, image, voice, video in one loop.

2. On-device & edge AI – privacy-first models running locally.

3. Agentic workflows – AI executes multi-step tasks, not just replies.

4. Built-in governance – bias, transparency, and safety as selling points.

5. 
   

Please note that AI space moves fast and there's a lot of new great products not covered in this lesson! I will update this periodically.