AI 101: December 2025 Edition

When I launched The Atomic Builder in January 2025, the standard AI advice was: use ChatGPT to summarise meeting notes, draft actions and emails…

Ten months later, AI can generate your entire slide deck, encode your team's workflows, and maintain living research hubs. These aren't theoretical capabilities—they're shipped features your competitors might already be using.

Here are five AI capabilities that either didn't exist or weren't ready for business use when we started 2025.

Each one comes with a 30-minute experiment you can run this week. By Friday, you'll know which ones actually work for your business.

1. NotebookLM Decks + Video: From document to slide deck in 30 seconds

What it is

Drop almost any source material into NotebookLM and get slide decks in different lengths, infographics with clean text and consistent styling, and video overviews that narrate your content, all images powered by Google's Nano Banana Pro image model.

Why it matters

Old way: Someone stays late manually cutting screenshots into PowerPoint.
New way: Every strategic document automatically becomes a slide deck and video explainer. Below is an explainer I made for this issue of the newsletter!

Try this today

Upload one asset (QBR pack, strategy memo, research report), then choose a visual style. In the notes, ask for: "A 5-slide board version” and “a 12-slide team version” with speaker notes. Or create a video explainer: "Generate a video for non-experts."

This is an example I made, pasting this newsletter content into NotebookLM (the video is hosted on X/Twitter).

2. Claude Skills: Workflows, not prompts

What it is

Skills are reusable folders of instructions and reference files that Claude loads when relevant. They turn vague prompting into repeatable workflows that run consistently across your team.

Why it matters

Old way: "Good prompts" in Notion that nobody uses.
New way: Your best workflows captured once, run consistently by anyone in your team, as and when Claude detects it needs to use them.

Think: "Our weekly product update" or "Client proposal format" as a Skill, not a template people break.

Try this today

Pick one recurring artifact you create (weekly update, client summary, meeting follow-up). Create a Skill with: tone/audience instructions, 1-2 example outputs. Next time, initiate the skill and watch Claude generate consistent output.

3. ChatGPT for Teams: Shared Projects + Group Chats

What it is

Shared Projects: Persistent workspaces (business / enterprise users only) where team members upload files, set instructions, and work from the same long-running context asynchronously.

Group Chats: a shared collective chat space for brainstorming with ChatGPT live, with AI mediating the discussion in real-time (can be used by users on a free plan).

Why it matters

Old way: Brilliant chats die in Slack screenshots!
New way: Persistent team workspaces for ongoing work; live AI-mediated sessions for decisions.

Projects solve the context handoff challenge - it’s always retained between, you, a colleagues and AI. Group Chats solve live alignment—instead of 20 minutes of "what does everyone think," AI surfaces trade-offs and helps teams decide faster.

Try this today

Project: (If you have the enterprise plan) Create a workspace for one initiative. Upload key docs, add instructions, share with 2 colleagues and work on the project together.

Group Chat: Invite 3-5 people to decide between a few options. Let ChatGPT surface criteria and trade-offs before the team votes.

4. Nano Banana Pro: Production-grade visuals

What it is

Google's Gemini 3 Pro image model (also known as Nano Banana Pro!), which is also integrated into NotebookLM (above). It can create clean, legible text in images, studio-quality designs in 1K-4K resolution, and consistent visual styles across multiple assets.

Why it matters

Old way: AI visuals that look interesting but unusable as company assets or board packs.
New way: Design-ready diagrams, mockups, and campaign assets generated where you work.

The constraint was always: AI images are fine for inspiration, but you can't ship them. Nano Banana Pro crosses that threshold. The decision flow below? Created with Nano Banana Pro in about 60 seconds. Clean text, professional layout, ready to use.

Try this today

This image model can be accessed from most of the suite of Google AI products.

For example, In Gemini, use the simple prompt "Beautify this slide" or "Create an infographic showing this process in 4 steps." to take an existing asset and improve it. Generate three variants. Save the best to your brand library.

5. Perplexity Spaces: Living research hubs

What it is

Dedicated workspaces in Perplexity where you upload files, set instructions, and query both your sources and the live web. They accumulate insights over time and can be turned into briefings when needed.

Why it matters

Old way: Every research question is a separate chat with no memory.
New way: A Space per topic that stores files, accumulates insights, and becomes institutional memory for you and your team.

Your competitive intelligence isn't a quarterly report someone forgets. It's a Space that stays current and answers new questions as they arise.

Try this today

Pick one theme (competitor, regulation, friction point). Create a Space. Upload 3-5 documents. Set instruction: "Track changes over time and highlight new information." Ask three real questions. Pin the best answer. Add your team to collaborate on the space.

Where to start (and how I can help)

Pick one capability from this list that solves real friction in your business—where you're currently paying someone to do repetitive production work that could be delegated.

Run an experiment from above. If the output is good enough to replace your existing process, you've just freed up capacity. If not, you've learned what "production-ready AI" looks like in your context.

Not sure which one to prioritise for your team, or need help running these experiments? I work with companies on exactly this - figuring out which capabilities map to your workflows and setting up those first successful runs. Reply to this email and we can talk through what makes sense.

Thinking bigger picture? If you're wondering whether your organisation is actually ready to adopt AI at scale - whether you have the strategy, governance, data infrastructure, and execution capability in place - that's what my soon to be released AI Readiness Audit is designed to answer. It's a 10-minute diagnostic that reveals where your gaps are across six critical dimensions, with clear scores and focused recommendations.

The companies that will lead in 2026 won't be the ones with the best AI strategy decks. They'll be the ones who moved fast - running tactical experiments like these while also building the organisational foundation to scale what works.

The Atomic Builder is written by Faisal Shariff and powered by Atomic Theory Consulting Ltd — helping organisations put AI transformation into practice.