AI Products: What to Expect in the Next 1-2 Years

by yaniskes

AI is no longer a feature — it’s becoming the product itself. In this post, I’ll share some fascinating facts about the AI industry’s growth and make projections on what AI-powered products we can expect to see in the next one to two years.

The numbers are hard to ignore. The global AI market was valued at $196 billion in 2023 and is projected to surpass $1.8 trillion by 2030 — a compound annual growth rate of over 37%. More telling: in 2024, over 72% of companies reported using AI in at least one business function, up from just 55% the year before. We are not in the early adopter phase anymore. AI has crossed the chasm.

Where We Are Today

AI technology

Let’s start with what’s already happening. Large language models have gone from research curiosity to infrastructure layer in under three years. OpenAI, Anthropic, Google, and Meta are in a full-scale arms race — not just on model quality, but on deployment cost, speed, and accessibility.

A few facts that stand out:

  • ChatGPT reached 100 million users in 2 months — faster than any consumer product in history, including TikTok (9 months) and Instagram (2.5 years).
  • GitHub Copilot users write code 55% faster on average, according to GitHub’s own studies. That’s not a marginal improvement — it’s a fundamental shift in developer productivity.
  • AI-generated content now accounts for an estimated 15% of all new web content, a figure that is growing every quarter.
  • The cost of running AI inference has dropped 99% in two years. What cost $100 to process in 2022 costs less than $1 today. This is the economic engine behind the explosion of AI products.

The Product Shift Nobody Is Talking About

AI product development

What excites me most is not the models themselves — it’s the product layer being built on top of them. We’re seeing a fundamental shift from AI as a feature (“our app has AI!”) to AI as the core product experience.

The traditional product development cycle was: discover → define → design → develop → test → launch. AI is collapsing this cycle. In some companies, entire product functions are being replaced or augmented by AI agents — not just assisted.

The most important product metric of 2025 is no longer DAU or retention. It’s task completion rate. Users don’t just want to be engaged — they want things done. AI products that actually complete tasks on behalf of users will win. Those that just assist will become commodities.

AI in the Enterprise: Already Reshaping How Companies Operate

Enterprise AI

Enterprise adoption of AI has accelerated dramatically. This is no longer pilot programs and proof-of-concepts — it’s production deployment at scale.

Microsoft has embedded Copilot across its entire Office 365 suite. Millions of enterprise users now draft emails, summarize meetings, generate presentations, and query company data through natural language — all without leaving their familiar tools. Early data shows a 70% productivity boost for repetitive knowledge work tasks.

Salesforce Einstein and HubSpot AI are rewriting what CRM means. AI now scores leads, drafts follow-up emails, predicts deal close rates, and auto-updates records from call transcripts. Sales teams that adopted these tools in 2024 report 30-40% increases in pipeline velocity.

ServiceNow is using AI to automate IT service management — resolving up to 80% of routine tickets without human intervention. The ROI is immediate and measurable, which is why enterprise IT budgets are shifting from headcount to AI tooling at an unprecedented rate.

Goldman Sachs, JPMorgan, and BlackRock are deploying AI for document analysis, contract review, and compliance monitoring — work that previously required armies of analysts. JPMorgan’s CoiN platform reviews commercial loan agreements in seconds versus 360,000 lawyer-hours per year previously.

The pattern is consistent: the highest-ROI enterprise AI use cases are the ones that eliminate the most repetitive, high-volume knowledge work. Document processing, customer support triage, code review, and data analysis are being transformed first. Strategic and creative work is next.

What’s Already Here vs. What’s Coming

Future AI products

As of April 2026, several of the trends I projected a year ago have already materialized. Let me be honest about what’s real now versus what’s still emerging:

Already mainstream by April 2026:

  • AI coding assistants — GitHub Copilot, Cursor, Claude Code, and Windsurf are now standard tools for professional developers. Cursor alone surpassed 1 million paid users. The debate is no longer “should I use AI for coding” but “which AI coding tool.”
  • AI personal assistants — OpenAI’s operator-style agents, Anthropic’s Claude with computer use, and Google’s Project Astra are real products in users’ hands today. They browse the web, fill forms, and complete multi-step tasks autonomously.
  • AI video generation — Sora (OpenAI), Runway Gen-3, and Kling AI produce near-photorealistic video from text prompts. Hollywood studios are already using these tools in production pipelines.
  • AI search — Perplexity AI has crossed 15 million daily active users. Google’s AI Overviews now appear on over 50% of searches. The way people find information has fundamentally changed.

Emerging in the next 12 months:

  • AI-native ERPs and business intelligence — Tools like Glean, Notion AI, and next-gen Tableau are making enterprise data queryable by anyone. “What were our top 5 churn reasons last quarter?” will be answered in seconds by anyone in the company, not just analysts.
  • Multimodal AI agents in physical spaces — AI that sees, hears, and responds in real environments. Retail associates powered by AI, AI-assisted surgical guidance, and autonomous quality control in manufacturing are moving from lab to floor.
  • AI-generated personalized learning — Platforms like Khan Academy’s Khanmigo and Duolingo Max are early examples. Expect fully personalized, adaptive learning experiences at scale across K-12, corporate training, and professional certifications.
  • Vertical AI agents — Purpose-built agents for specific industries: Harvey AI for legal, Abridge for medical notes, Synthesia for corporate video, and dozens more. The vertical SaaS market is being rebuilt from scratch around AI-native workflows.

The Risk PMs Must Not Ignore

AI risks and challenges

Not everything about AI in products is upside. We need to be honest about the risks building into our roadmaps right now.

Trust and reliability remain the biggest unsolved problem. AI systems hallucinate. They make confident mistakes. Building a product where a user delegates a task to an AI — and the AI does something wrong — requires entirely new UX patterns for error recovery, confidence signaling, and human override.

Dependency risk is underrated. If your product’s core value is powered by a third-party model, you are exposed to pricing changes, API deprecations, and model behavior shifts. The companies building proprietary fine-tuned models on their own data are building real moats. Those just wrapping GPT-4 are building on sand.

Regulatory headwinds are coming. The EU AI Act is already in force. US regulation is slow but inevitable. Products in healthcare, finance, and HR are particularly exposed. Build with compliance in mind from day one — retrofitting it is expensive and painful.

Final Thought

The AI industry is not in a bubble — it’s in a compression. The gap between what AI can do and what products actually deliver to users is still enormous. The best product opportunities over the next two years are not in building better models. They’re in building better products around existing models — with great UX, deep domain expertise, and real workflows that replace real work.

Related Posts

Leave a Comment