AI Tools 2026
The AI stack we use every day
By Alexander Schneider, founder of OOTO — Out of the Ordinary
New AI tools are released every week. Every week, someone promises that this particular tool will change everything. And every week, entrepreneurs face the same dilemma: Which one is really worth it?
We don’t answer this question in theory. We’ll show you exactly which tools we use every day here at OOTO—and why. No affiliate marketing, no sponsored recommendations. Just what works for us in practice.
First, an important note: AI tools are no substitute for strategy or subject matter expertise. They are amplifiers. A good tool in the hands of someone without experience produces mediocre results more quickly. A good tool in the hands of someone with expertise produces excellent results that would have taken hours to achieve without AI.
With that in mind, here is our stack.
Large Language Models: What You Need to Know Before You Get Started
LLM stands for Large Language Model. These are language models that have been trained to understand and generate text and solve problems. ChatGPT, Claude, and Gemini are all LLMs that have been made accessible via a chat interface.
The big question isn't, "Which one is the best?" The right question is, "Which one is suitable for which use case?" There is no one-size-fits-all solution. There is meaningful specialization.
As of March 2026, there are three dominant providers:
• Anthropic with Claude (Opus 4.6, Sonnet 4.6, Haiku 4.5)
• Google with Gemini (Gemini 3 Pro, Gemini 3 Flash)
• OpenAI with ChatGPT (GPT-5.2 and variants)
In addition, there’s Perplexity as a research layer and specialized tools for specific tasks. Let’s go through them one by one.
Claude: Our main tool for everything related to language and thinking
Claude by Anthropic is by far the LLM we use most often. Not because it ranks first in every single benchmark, but because it works best for our work.
Why Claude?
Claude was developed using a different design approach than most other LLMs. Anthropic places great emphasis on what they call “Constitutional AI”: the model is intended not only to be helpful, but also reliable and honest. In practice, this means that Claude hallucinates less often, admits to uncertainty more frequently, and produces outputs that are closer to what you actually need.
In our view, Claude is the most powerful model on the market for writing and document work. Claude Sonnet 4.6 supports a context window of up to one million tokens, which is equivalent to approximately 750,000 words or an entire book manuscript in a single conversation. This makes it the ideal tool for long documents, complex briefings, or extensive content projects.
Claude Sonnet 4.6 dominates the SWE-bench (coding benchmark) with a score of 72.7%, placing it well ahead of GPT (54.6%) and Gemini (63.8%). This holds true in practice as well: Cursor, the leading AI coding tool, describes Claude as “state-of-the-art” when it comes to understanding complex codebases.
How we use Claude at OOTO
Blog posts and long-form content (from research and structuring to the final polish)
Customer documents, reports, and proposals
Strategy Brainstorming and Concept Development
Analysis of uploaded documents (PDFs, Word files, data)
Technical Implementation and Prompting Frameworks
Our tip: Claude isn't just a writing assistant. Treat it like a thinking partner with a wealth of knowledge and a keen focus on you. The more context you provide, the better the results.
Available as: Free version (claude.ai), Pro plan for ~22 CHF/month (claude.ai/pro).
Gemini: The search engine powered by Google
Gemini is Google's answer to ChatGPT and Claude. And it has turned out to be more powerful than many expected. Gemini 2.5 Pro, in particular, has established itself as a serious alternative.
What Gemini Does Better Than the Competition
Gemini has a structural advantage: it is deeply integrated into the Google ecosystem. Google Search, Google Workspace, YouTube—all of these feed into Gemini’s capabilities. For research tasks that rely on up-to-date web information, this is a significant advantage.
Gemini 2.5 Pro has the largest context window in the comparison and is also the most cost-effective model in the high-end category. This is particularly relevant for large volumes of documents or data-intensive analyses.
In independent benchmarks (Artificial Intelligence Index), Gemini 3.1 Pro ranks at the top, demonstrating that Google is continuing to close the gap with Anthropic and OpenAI.
How we use Gemini at OOTO
Market research and competitive analysis with up-to-date web access
Summarize large volumes of documents thanks to a large context window
Research Tasks with Google Workspace Integration
First drafts for content formats where speed takes precedence over depth
Gemini isn't a copy of ChatGPT. It's Google's tool, excelling in everything related to up-to-date web information and the Google ecosystem.
Available from: Free version (gemini.google.com), Google AI Pro for ~22 CHF/month.
ChatGPT: The pioneer that remains a creative sparring partner
ChatGPT was the first large language model (LLM) to reach the mass market. It defined the category. And it remains the tool with the largest user base worldwide, with over 300 million active users per week (OpenAI, as of early 2025).
To be honest: In our daily workflow, ChatGPT has ceded its spot as our primary tool to Claude. Not because GPT-5.2 is bad (quite the opposite, in fact), but simply because Claude is a better fit for our use cases.
Where ChatGPT continues to excel
According to independent tests, GPT-5.2 excels at complex reasoning and multimodal support (the simultaneous processing of different input formats). For complex, multi-step reasoning tasks and scenarios where you combine different media types, ChatGPT is an excellent choice.
In our experience, ChatGPT is particularly useful when you need creative ideas that come from a different angle. As a sounding board for concepts, campaigns, or positioning strategies, it offers insights that differ from those of Claude. That’s valuable.
How we use ChatGPT at OOTO
Creative brainstorming sessions and positioning sparring
Multimodal tasks (combining text, images, and data)
Second opinion on the concepts we developed with Claude
Occasional reasoning tasks that require thinking through multiple steps at the same time
ChatGPT isn't a worse tool than Claude. It's a different tool. Using two different LLMs with slightly different strengths isn't inefficient; it's quality assurance.
Available as: Free version (chat.openai.com), Plus plan for ~22 CHF/month.
Perplexity: The research engine with actual source citations
Perplexity is not a traditional LLM interface. It is an “answer engine”—a hybrid of AI reasoning and real-time web search—that always provides answers with cited sources.
That is the key difference: ChatGPT and Claude primarily operate based on their training data. Perplexity searches the web in real time and synthesizes the results. For time-sensitive information (current market data, the latest studies, yesterday’s news), this is a fundamental advantage.
Perplexity now processes over 780 million search queries per month (as of May 2025), boasting one of the fastest growth rates in the entire AI sector.
The Deep Research Feature
The feature that makes Perplexity particularly appealing to professional users is called Deep Research. Instead of providing a single answer, the system performs 20 to 50 iterative search steps, evaluates sources, and synthesizes a multi-page research report—all in two to four minutes. This is something that would take a human researcher hours to complete.
In a benchmark (Humanity's Last Exam), Perplexity Deep Research achieves an accuracy rate of 21.1%, outperforming Gemini Thinking, o3-mini, and DeepSeek-R1. For SimpleQA factuality tests, the accuracy is 93.9%.
How we use Perplexity at OOTO
Market and Competitive Research Using Up-to-Date Data
Quick fact-checking during content creation
In-depth research for client briefings and strategy documents
Industry trends and new developments in the field of AI
Perplexity isn't a replacement for Claude. It's the tool you use before working with Claude to ensure that your starting point is up-to-date and accurate.
Available as: Free version with daily limits, Pro for ~$22/month (unlimited).
NotebookLM: The best way to make sense of large volumes of documents
Google NotebookLM is the most underrated tool on this list. It’s not a general-purpose LLM interface. It’s a specialized research tool that works exclusively based on the documents you upload.
Here’s how it works: You upload sources (PDFs, Word documents, Google Docs, URLs, YouTube videos, even audio files). NotebookLM analyzes these sources and answers your questions based solely on the material you’ve provided. No hallucinations from the general training data pool. Just your material, structured and made accessible.
Since December 2025, NotebookLM has been running on Gemini 3, with significantly improved reasoning and multimodal understanding. As a result, the outputs are more accurate, especially when dealing with dense, complex texts.
What NotebookLM Can Do
Audio Overviews: Converts documents into podcast-style conversations (perfect for listening to long reports on the go)
Video Overviews: Generates visual video summaries with AI narration
Mind Maps: Interactive visual representations of complex relationships
Slide Decks: Create presentations directly from your sources
Data Tables: Structures information from text into clean tables (new as of December 2025)
Deep Research: An active web research agent that builds bibliographies on its own
How we use NotebookLM at OOTO
Analysis of extensive client documents and briefings
A synthesis of several market studies or reports
Quick familiarization with unfamiliar fields
Audio summaries for lengthy research reports on the go
Convert meeting transcripts into structured action item tables
NotebookLM has a clear limitation: it works only with the data you upload. This is also its greatest strength. No hallucinations, since the model doesn’t invent information that isn’t in your sources.
Available from: Free at notebooklm.google.com. NotebookLM Plus as part of the Google One AI Premium Plan (~CHF 22/month).
Nano Banana: Google's image-generation tool that went viral in August 2025
Nano Banana is Google's AI image generation model, part of the Gemini ecosystem. The name may sound playful, but the model is anything but. As of 2026, it is one of the most powerful image generation solutions on the market.
Here’s a quick recap: The original Nano Banana (Gemini 2.5 Flash Image) was released in August 2025 and went viral in a way no one had anticipated (especially the trend of turning selfies into 3D figures). In November 2025, Nano Banana Pro followed, built on Gemini 3 Pro, offering studio-quality results and precise text rendering. In February 2026, Nano Banana 2 arrived, combining the quality of the Pro model with the speed of the Flash model.
What Nano Banana Can Do
Text-to-image generation in up to 4K resolution
Precise adherence to instructions for complex compositions
Consistency across multiple characters and objects
Accurate text rendering in images (text in designs, mockups, posters)
Real-time Web Grounding: Nano Banana 2 retrieves up-to-date information from Google Search to generate contextually accurate images
SynthID Watermark for Transparent AI Marking
How we use Nano Banana at OOTO
Visuals for LinkedIn posts and blog headers
Mockups for client presentations
Social media content without stock photo restrictions
Quick concept visualizations for client meetings
Important: Nano Banana isn't a competitor to Midjourney in the traditional sense. It's an integrated Google tool that's most useful if you're already working within the Gemini ecosystem and want to generate images directly within your workflow.
Available starting: Free in the Gemini app (with limits). Nano Banana Pro for Google AI Pro/Ultra subscribers.
The OOTO AI Stack: How the Tools Work Together
Knowing individual tools is one thing. Understanding how they work together in a workflow is another. Here’s what our typical workflow looks like:
Workflow 1: Create a blog post
1. Perplexity:
Research current data, benchmarks, and new developments on the topic. Review sources.
2. NotebookLM:
Upload relevant PDFs, studies, or reference materials. Generate a briefing document.
3. Claude:
Write the full article using the briefing from NotebookLM as a reference. Ensure it complies with E-E-A-T guidelines, includes citations, and maintains the OOTO tone.
4. Nano Banana:
Create a header image and social media visuals.
Workflow 2: Client Research and Strategy Briefing
1. Perplexity Deep Research:
Market, competitors, and industry trends are automatically synthesized into a structured report.
2. NotebookLM:
Analyze customer documents and the Perplexity Report together. Identify key insights and correlations.
3. Claude:
Draft a strategy paper, proposal, or client presentation.
Workflow 3: Creative Concept
1. ChatGPT:
Creative brainstorming: positioning ideas, campaign concepts, unconventional approaches.
2. Claude:
Develop the best concept, refine it, and turn it into a concrete document.
3. Gemini + Nano Banana:
Integrate current market data and generate visual concepts.
What we don't use and why
Just as important as the recommendations: what we deliberately left out of our stack.
Midjourney: Excellent image quality, but no web access, no text integration, and the subscription model has lost value compared to Nano Banana Pro.
Jasper, Copy.ai, and similar services: These are front-end interfaces built on the same models (GPT, Claude) that you can use directly at a lower cost. You’re paying for templates, not for better AI.
Grok (xAI): More affordable , growing rapidly, but doesn't yet offer a compelling use case compared to our existing marketing and strategy stack.
DeepSeek: Powerful and cost-effective, but privacy concerns (servers located in China) make it unsuitable for customer data.
What you, as an SME business owner, should do right now
If you don't have an AI stack yet, here are the priorities:
Week 1: Subscribe to Claude Pro. Start using it to complete daily writing and thinking exercises. Provide as much context as possible.
Week 2: Add Perplexity Pro. Use it for every research task before switching to Claude.
Week 3: Try NotebookLM for free. Upload your most important customer documents and experiment with audio overviews and data tables.
Month 2: Google AI Pro for Gemini and Nano Banana, giving you access to virtually the entire Google AI ecosystem.
Total cost: about CHF 60–80 per month. For everything. That’s roughly equivalent to one hour of consulting time. The return: hours of work saved every day, better results, and well-informed decisions.
This stack is not complete, and that is intentional
This article focuses on large language models and the tools we use every day. However, the AI ecosystem is much broader. There is a growing category of specialized tools that handle specific tasks and integrate seamlessly into a modern workflow.
Here’s an example we use regularly ourselves: Plaud.ai. It’s an AI-powered meeting tool that records and transcribes conversations and immediately generates structured summaries, action items, and draft follow-ups. Anyone who has a lot of meetings knows how much time is spent on follow-up work afterward. Plaud solves exactly that problem.
Other categories that may be relevant depending on the company:
Meeting Transcription and Summary (Plaud.ai, Otter.ai, Fireflies)
Automation and Workflow Integration (n8n, Make, Zapier with AI triggers)
Specialized image and video generation (Runway, Kling, Adobe Firefly)
Coding assistants (Cursor, GitHub Copilot, Claude Code)
Presentation and Document Generation (Gamma, Canva AI)
The choice always depends on the specific use case. Our advice: First, identify which recurring tasks in your business take up the most time. Then check if there’s a specialized tool that addresses exactly that. Not everyone needs everything. A lean, carefully curated stack is better than a bloated one with ten tools, five of which are rarely even opened.
Conclusion: A good stack isn't a cost center—it's a lever
Some entrepreneurs dismiss AI tools as mere toys, while others exaggerate their benefits as a panacea. Both views are wrong.
A well-designed AI stack is a productivity booster. It doesn’t replace expertise—it amplifies it. Those who know what they want and which tool is best suited for the task work faster, better, and with greater confidence than was imaginable just two years ago.
This isn't just hype. This is our everyday life.
Want to know what an AI stack would look like for your small business?
Every company has different processes, different teams, and different bottlenecks. We help you define the right tech stack and show you how to integrate it into your daily workflow. No tool hype—just a concrete focus on your ROI.
Book a strategy consultation now → strategie.ootolab.com
Frequently asked questions
Which AI tool is best for beginners?
Claude. It’s the most user-friendly, consistently delivers high-quality results, and feels like a natural conversation. Start there and expand your tech stack once you know what you need.
Do I really have to subscribe to multiple tools?
No. You can start out with just Claude and get very far. The real value of the combination only becomes apparent when you work systematically. For solopreneurs or small teams, Claude and Perplexity are often all you need.
How long does it take to implement an AI stack?
The tools themselves can be set up in minutes. The real time investment lies in developing good prompts and understanding when to use which tool. Expect it to take two to four weeks before the workflow feels natural.
Is the data secure when using these tools?
That depends on the provider and the plan. With Claude Pro, Gemini, and Perplexity Pro, data is not used for model training by default. For highly sensitive customer data, we recommend reviewing the privacy policies of the respective providers and using API access with appropriate data protection agreements.
How much does a complete AI stack cost per month?
Claude Pro (~CHF 22), Perplexity Pro (~CHF 22), Google AI Pro with Gemini + NotebookLM Plus + Nano Banana Pro (~CHF 22): A total of about CHF 60–70 per month for the full stack.
Sources
Anthropic, Claude 4 Product Page (2025) — anthropic.com
Artificial Intelligence Analysis Index (2026) — artificialanalysis.ai
Aggarwal et al., “GEO: Generative Engine Optimization” — Princeton / Georgia Tech (2023) — arxiv.org
Perplexity AI, Introducing Deep Research (2025) — perplexity.ai
Perplexity AI Features & Statistics 2026 — index.dev
Google DeepMind, Nano Banana Pro (November 2025) — blog.google
Google DeepMind, Nano Banana 2 (February 2026) — blog.google
Google Workspace Updates, Data Tables in NotebookLM (December 2025) — workspaceupdates.googleblog.com
Tom's Guide, ChatGPT vs. Claude (March 2026) — tomsguide.com
OpenAI, ChatGPT Usage Statistics (2025) — openai.com
