Reid Hoffman on using AI to adapt, earn more, and stay relevant

Executive overview

Most people using AI are barely scratching the surface. Reid Hoffman argues the window to adapt is closing fast — within a few years, there will be no such thing as an individual contributor who doesn't work with a suite of AI agents.

The opportunity is immediate: get proficient with AI in a specific domain, make that proficiency visible, and go help companies transform. Small businesses that adopt AI will have a structural advantage over large incumbents. Those that don't will struggle.

Build AI fluency as a reflex — before every task, ask how AI could help.

From easy to advanced: levels of AI use

• Easy: use AI chatbots substantively, not just for quick lookups — think out loud, voice-prompt, ask follow-up questions
• Use voice prompting: speaking is faster and produces richer input than typing seven words
• Ask the AI to write you the right prompt before you run it — the resulting two-page prompt gets far better results
• Medium: assign AI agents persistent roles in ongoing workflows, not one-off tasks
• Advanced: add a meta-agent that synthesises across all your agents and pulls in external data — other creators, competitors, adjacent industries
• Always prompt AI to do live web research for current topics — its training data is 18 months out of date

Role-based prompting

• Ask AI to answer a question from multiple perspectives: technologist, investor, policy-maker, safety expert
• Then ask: "Are there roles I haven't thought of?"
• Use contrarian prompting: ask the AI to argue against your position, then ask what you're missing to argue for it
• This reveals blind spots and the full landscape of a decision

How to double income from a 9-to-5

• Companies know they need AI transformation and are actively searching for people who demonstrate AI fluency
• Get proficient in AI applied to a specific domain — finance, marketing, supply chain, sales — and make it findable (LinkedIn, social media)
• The lucrative jobs aren't only AI researchers; applied AI practitioners will be in demand across every function
• Pattern: build proficiency in a domain, demonstrate it publicly, go help other businesses implement it

The future of B2B software

• SaaS moats were built on switching costs and feature accumulation — that is now under pressure
• With AI coding, companies can build and maintain custom software more cheaply than buying off-the-shelf
• Software engineers won't disappear — they'll shift from writing code to conducting AI coding agents
• The human-plus-AI combo will outperform AI alone for a while; humans provide context and judgment that agents can't walk around and observe

Small business vs large business in the AI era

• Small businesses can adapt faster than large incumbents locked into industrial-efficiency models
• Expect a flood of AI-generated content — but persistent demand for human-created and group experiences will remain
• Trust is an asymmetric advantage: solo AI experiences will dominate, so group and community formats become differentiators
• If your business will be commoditised (e.g. test prep), rebase on AI and ask: what can I add that a solo AI session can't? — group experiences, community, accountability, personal brand

The future of invention

• 60–70% probability: most future inventions over the next 50–100 years will be human-plus-AI
• ~25–30%: primarily AI-driven, with humans cross-checking
• ~5%: purely unassisted human insight
• Already happening: physicists using AI to solve problems that humans couldn't reach alone

One thing to do before February 2027

• Before every task — vacation planning, writing, investing, difficult conversations — pause and ask: how could AI help here?
• You won't always act on it, but building that reflex is the core skill
• The people who treat AI as optional will be outpaced by those who treat it as the default starting point