50 AI tools you should be using right now, plus 4 skills you need to thrive in an AI-driven future

Thought of the week: how to survive when machines are smarter than us

I want to unpack a fascinating blog that Sam Altman posted last week about the future of AI and what a post-AGI world will look like; especially given the tweets by his team that their upcoming ChatGPT 4.5 model shows signs of AGI. My key takeaways from his blog can be split into two key themes: i) how AI infrastructure and value is mushrooming and ii) the key skills we all need once we have super-duper AI.

On building AI infrastructure and getting economic value from those pipes, he argues that:

1. The more you spend, the more you get (scaling): Doubling compute budgets yields consistent improvements in model capability.
2. AI will only get cheaper, which will drive widespread adoption (e.g. the cost/token for GPT-4 has fallen 150x in the last 12 months). This is triggering the Jevons paradox: the cheaper something gets, the more people use it.
3. The economic value of AI is super-exponential: From coding apps in days rather than months to similarly compressed AI-driven drug discovery, AI tools are already giving users way more value in terms of saved time/effort/cost than what they are spending on it.

None of the above is controversial, but it does confirm that we are at the very beginning of an explosion in AI capability and, critically, deployment. So what does Sam think we should be doing about it?

The key skills we’ll all need to survive

Even though everyone (who can afford it) will have access to what I call 3c (conversational, context-aware and customised) AI agents and systems, this will in no way level the playing field. In fact, the divide will shift from AI 'haves' and 'have-nots' to those who can pay for the best AI agents but also have the right mindset. For Altman, this mindset will be:

1. Critical thinking. The intelligence to ask the right questions, question the wrong answers or, more importantly, get the AI to do the right things. For example, I can sign up to Instagram and tacitly choose to allow Meta to choose how I waste my precious time on this earth or I can direct an AI to teach me stuff that makes me smarter.
2. Economic and social agency. AI is NOT a social leveller. It will make the world more unequal as it massively favours capital over labour. Your ability to ‘succeed’ will still be down to access to the economic and social networks that keep rich people rich and poor people poor.
3. Wilfulness and resilience. This is especially important in any era of rapid change. If you’re scared of change or wilt at the slightest resistance to your brilliant idea, you will fail to maximise the opportunities presented by AI.
4. Never stop having ideas. For me, this is critical. AI enables anyone to prototype an idea, whether it’s your first novel or the next big thing in tech quickly and at minimal cost. Equally cheaply, it can give you a much-needed second opinion when you need to kick the tyres of an idea.

In summary, Big Tech and Big Government are now hell-bent on ‘mainlining’ AI into our lives whether we like it or not. Resistance is futile, so lean in but don’t forget that it’s the soft skills that will mark you out from those that rely only on AI.

I often get asked to recommend an AI tool to do XYZ, so I’ve compiled a useful list of the (almost) 50 tools that I’ve played with over the last 12 months to ease you into an AI-augmented workflow.

When you’re too lazy to write

• Claude.ai (Free / Pro $20/mo) – Though Anthropic has yet to release a true 'thinking' model, Claude 3.5 still writes the most natural-sounding AI-generated text. If you're not a writer, this is the best tool, especially when using the option to write in a particular style.
• ChatGPT (Free / Plus $20/mo) – GPT-4o (and even GPT-3o mini) can be fine-tuned to match your writing style. Customise its output with a negative prompt to refine tone and style e.g. "Avoid the word 'delve'. Do not use complex or abstract terms such as 'meticulous,' 'navigating,' 'complexities,' 'realm,' 'bespoke,' 'tailored,' 'towards,' 'underpins,' 'ever-changing,' 'ever-evolving."
• Gemini, DeepSeek, Qwen – All good enough for basic writing tasks.

Getting AI to do your homework for you

• OpenAI Deep Research – Provides detailed, academically rigorous reports for $200/month through ChatGPT Pro.
• Perplexity AI Deep Research – Offers fast, comprehensive reports for free (five queries/day) or $20/month for up to 500 queries.
• DeepSeek’s R1 model in DeepThink mode – A free alternative to Deep Research but limited by Chinese guardrails.
• Google Deep Research – Available as part of the Google One “AI Premium” plan for approximately $19.99/month, integrating with Google Workspace tools.
• Google’s Learn About – A free experimental tool for exploring any topic.

For those who hated art class

• Visual Electric (Free / paid plans TBA) – Truly image generation for dummies, with excellent results and minimal prompt engineering. Less prone to excessive content guardrails than competitors.
• Midjourney ($10–$60/mo) – Still the best for artistic and highly detailed AI-generated visuals.
• Runway ML (Free / Paid plans start at $12/mo) – AI-powered image and video editing, with strong inpainting and text-to-image capabilities.
• Stable Diffusion (Open Source) – Free and customisable AI image generator that runs locally.

Automating the ever-so-dull process of creating presentations and trying to liven them up with nice infographics

• Beautiful.ai ($12–$40/mo) – AI-powered slide creation with Smart Narratives that automatically highlight key data. You can even integrate it with Zoom to turn meeting transcripts into slide decks.
• Gamma (Free / Pro $10/mo) – Generates interactive presentations with Live Data Tiles (e.g. real-time Salesforce metrics) and embedded polls.
• Napkin – A genius little tool that uses AI to create simple but beautiful explanatory infographics. Example above.

If you’ve always wanted to make your own pop music but have no discernible talent

• Suno AI (Free / Paid plans TBA) – Fully AI-generated songs, including lyrics and vocals.
• Udio (Free / Pro plans TBA) – Capable of generating inoffensive AI tracks complete with lyrics. Sadly for musicians, probably the near-future of music creation.
• AIVA (Free / Pro $15/mo) – AI-composed orchestral scores exportable as MIDI files, commonly used for game and film soundtracks.
• Mubert (Open-source API) – AI-generated music for streaming, film and interactive experiences.

When you have time (and money) to waste on creating uncanny AI Valley videos

• Runway ML (Free / Pro plans start at $12/mo) – AI-powered video editing and generation. Their cartoon generation feature is incredible.
• Pika Labs (Free / Pro plans TBA) – Converts text to video with cinematic visuals.
• Hailuo – AI-powered video generation.
• Kling – Text-to-video AI with realistic motion and speech synthesis.
• Veo2 by Google – As yet unreleased AI-powered video generation from Google, but widely believed to be the best.
• SkyReels – The brand-new kid on the block, a free, open-source and GenAI platform.

When you can’t be everywhere at the same time (Cloning yourself)

Voice cloning

• ElevenLabs (Paid) – Hands down the best high-quality AI voice cloning.
• Voice Pro – A free, open-source alternative for AI voice cloning, but you need a powerful computer and sound-coding knowledge to download it and set it up.

Video cloning

• HeyGen (Pro $29/mo) – AI-powered avatar video creation.
• Synthesia ($30/mo) – Over 150 AI avatars with emotion control and scriptwriting integration.
• DeepBrain AI – Realistic AI avatars for presentations and explainers.
• D-ID – AI video generation from images and text, great for deepfake-style avatars.

Coding assistants for people who can’t code

• Bolt.new – A brilliant tool for creating an app’s front end (and back end with a bit of effort). If you want to see the future of coding then play with Bolt and you’ll soon understand why we (or our agents) will all be ‘AI builders’.
• Loveable.ai – Similar to Bolt, Loveable lets you code a simple complete app. Free for up to five messages per day and then $20/month thereafter.
• Replit – Similar to Bolt and Loveable but can be used on a mobile device.

When you want to geek out (and have a lot of time to waste) playing with local LLMs

• Ollama (Free) – The easiest way to run AI models like Mistral and LLaMA locally.
• LM Studio – A desktop app for running local LLMs without complex set-up.
• GPT4All– Open-source chatbot that runs locally.

When you’re ok letting an AI control your browser or computer

• Operator ($200/mo) – Slow, but it just works out of the box. This will soon be free, so only pay for it if you're impatient. Current use cases include filling in forms.
• Browser-use – Totally free open-source version of Operator.
• Anthropic's Computer Use ($50/month Claude Pro subscription)
• Proxy – Another brand-new tool that wants to do your browsing for you.

Next week, I’ll be covering the best AI meeting bots, job search tools and much more.

What we’re reading this week

• Elon Musk launched Grok 3, typically claiming it's the best LLM available and is "maximally truth-seeking", whatever that means!
• Google just launched an AI co-scientist, a multi-agent research assistant (built on Gemini 2.0) that accelerates scientific discoveries by generating and validating new hypotheses across areas such as medicine and genetics.
• Perplexity open-sourced R1 1776, a retrained version of DeepSeek’s reasoning model that delivers the same performance without built-in censorship.
• Meta announced Project Waterworth, an ambitious initiative to lay the world’s longest sub-sea internet cable.

That's all for this week! Subscribe for the latest innovations and developments with AI.

So you don't miss a thing, follow our Instagram and X pages for more creative content and insights into our work and what we do.