Aller au contenu principal

AI for Building an AI-Native Business Webinar April 2nd

•1:13:41
AI Tool Review

A tour of my AI-native business stack: Claude Code as the command center, a clean Notion knowledge base, Unipile for LinkedIn data, Modal to run automations online, and Gemini for video analysis. It includes live builds you can copy.

Pour qui: Business owners, solo operators and AI implementers who want to connect their business data to Claude Code and run real automations for themselves or their clients.

Téléchargement gratuitClaude Code + Unipile + Modal Prompt PackThe prompts I gave Claude Code in the webinar to sync LinkedIn posts to Notion, deploy to Modal, analyze video with Gemini and automate LinkedIn outreach.
Télécharger

À retenir

  1. Build software from Lego blocks

    I share a friend's idea, the "software factory". You used to choose between a SaaS product (it scales but is hard to customize) and custom consulting (tailored but impossible to repeat). Now you keep small building blocks, such as social tracking, lead qualification and reporting over WhatsApp, Slack or email, and ask AI to put them together. You get workflows that are both customizable and repeatable.

  2. Your knowledge base comes first

    Claude Code needs one source of truth for your business: onboarding documents, client transcripts, CRM, testimonials and content. This can be a Notion workspace, a Google Drive or plain CSV files. Ask Claude Code to map every database and how they connect, then reorganize them logically. In my experience, this cleanup takes 80 to 90% of the time. The first pass is always done by a human, and automation comes after.

  3. n8n and Make still have a place

    I disagree that no-code is dead. People think visually, so n8n and Make.com workflows are easier to share with a team and document, for example to change a prompt, an API key or a model. Claude Code is better for solo operators, but it cannot run automations by itself. You pair it with a service that hosts and runs your scripts.

  4. Claude Code + Unipile + Modal

    My content database refreshes every morning using three pieces. Unipile connects social accounts (LinkedIn, WhatsApp, Instagram, X, Telegram) through an API. Claude Code writes the script. Modal puts Python scripts on the web and runs them on a schedule. Modal's free tier covers my use, since I spend less than the $5 monthly credit. I estimate you can build this in fifteen minutes.

  5. Use Gemini to analyze video

    When you call Gemini through its API, it can analyze a video and produce a transcript with accurate timestamps, with no separate transcription tool. You can use it to tag a whole video library, for example finding every clip where a woman talks about physics. In one test, Gemini solved advanced equations I had written on paper and held up to the camera.

  6. Pay for new data, not data handling

    I suggest paying for services that give you new data rather than apps that rework data you already have. Unipile is worth paying for because it opens up data locked in your LinkedIn account. With Claude Code you can build the logic around it yourself. I dropped PhantomBuster because Modal with Unipile was simpler and easier to customize.

  7. One interface for all your models

    I store API keys for Gemini, Anthropic and the other major LLM providers inside my Claude Code workspace. When I ask Claude Code to use Gemini, it calls the Gemini API directly. That gives me one place to run my whole business, and I can pick whichever model suits the task. For example, I pick Gemini for documents, video and audio.

Outils présentés

Questions du public

What exactly do you mean by a knowledge base?

It is a fancy name for something simple: a Notion workspace that holds my whole business, including the competitors database, leads databases and CRM. It is not about technology. It can be a set of documents on Google Drive, or just CSV files inside your Claude Code workspace, and you would still get a lot of value. What matters is that it is the single source of truth and kept clean.

If I build this for a client, how do onboarding and handoff work?

Book a thirty-minute call and create the accounts together, for example Modal and Unipile. With access to those accounts, you connect Claude Code and ask it for the workflow you need, such as fetching someone's LinkedIn posts or connecting with people from a Google Sheet every twenty minutes or every few hours. Deploy it to Modal, then disconnect your API access and hand the setup over to the client.

How do you handle data security and privacy?

For Claude Code, you can ask Anthropic's sales team for zero data retention so your data is not kept, and turn off the "help improve Claude" setting. Google says it does not use your data by default for Gemini, but I do not fully trust that. I mostly work with public information. For clients with a lot of private data, I suggest local models, which need a computer that runs them locally.

Should I use Claude Code or n8n for a client project?

I now use Claude Code for everything and have stopped using n8n, but it depends on the client. If the client does not want to learn AI or debug things themselves, use Claude Code with Modal. If a team needs to see, share and edit workflows, the visual tools n8n and Make.com are still the better choice.

À regarder ensuite

Webinaires associés

Implémentation: Des systèmes IA intégrés à vos process→