Based on my analysis of current AI developments worldwide, here are six trends I anticipate emerging over the next 6-18 months (the pace of evolution is extraordinary).
1) Applications with Dynamic User Interface
AI models will adapt user interfaces to match specific work environments, individual worker roles, and personal preferences.
This represents a fundamental disruption to traditional SaaS, driven by the rapid development capabilities of coding agents and AI systems’ ability to integrate data through standardized interoperable interfaces (API and MCP).
Note: The building blocks already exist.
2) Personal AI Assistant in a Box
We’ll soon see laptops and mini computers (Mac Mini, Raspberry Pi, Mini PCs) pre-configured with local AI agents.
Cloud-based offerings from hyperscalers will follow the same pattern.
3) Specialized Workforce as a Service
Specialized AI workers with domain-specific skills will be available on-demand to excel at particular tasks within given cultural and regulatory environments.
This represents the evolution of SaaS in the coming disruption.
For example, you would expect Salesforce to launch a full-fledge Sales Agent Worker as a Service (Agentforce is the first iteration).
Signals: Claude Cowork with plugins (such as Legal), released recently, and OpenAI’s Frontier, just announced, are clear steps in this direction.
4) Enforced Agent Identity
We’re moving toward a world of 9 billion people and potentially 90 billion agents.
We’ll need to establish ownership and representation for each agent to manage critical flows—particularly financial transactions (banks should prepare for autonomous personal AI agents conducting banking operations) and legal representation for actions and decisions.
Think Security, Security, and Security.
5) Front-End APIs (App-to-App APIs)
Applications running on front-ends and mobile devices will require APIs to integrate and communicate securely within trusted environments on smartphones, laptops, and other devices, with compute happening at the edge only.
This shift is supported by the rebalancing we’re witnessing: while centralized cloud data centers continue to grow, we’re simultaneously seeing more powerful smartphones and laptops equipped with GPU AI computing capabilities (like Apple’s M series).
6) Portable Context Database
Your context is essential for a genuine agent experience. If you haven’t tried ChatGPT with memory activated, you’re missing the true potential of AI assistance-and this is just a preview.
Letta AI and OpenClaw (previously known as Clawdbot) offer long-term memory mechanisms, but these contexts remain constrained to specific AI technologies/ models. I anticipate a standard, enabling your context and credentials to flow seamlessly between Anthropic Claude, ChatGPT, Google Gemini, and Microsoft Copilot, with clear separation between:
- Personal Private Context
- Personal Group Context
- Personal Public Context
- Work Private Context
- Work Corporate Context
- Work Secret Context.
What This Means
These trends enable the development of robust business strategies, technology roadmaps, solutions, consulting services, and vertical startups.
Most importantly, they help protect your people during continuous and accelerating changes in how we work.
Whatever you’re planning, consider how it fits your business and think sovereign, whether that’s on-premise AI infrastructure (hello Mistral AI), regional data residency, or proprietary context databases; this is your competitive moat.
Meanwhile, major AI players are absorbing business value like black holes. Remember: regulatory frameworks may slow execution, but they also protect your business by limiting market penetration from external competitors.
Your most important assets are the business experiences you offer, your staff, and your data. AI is coming for everything in between.
Yannick HUCHARD
