← Back to search
10 Free AI Tools I Actually Use (And What I Built With Each)
AI News Today | Julian Goldie Podcast · 2026-11-10 · 11 min
Show full episode description
10 Free AI Tools I Actually Use (And What I Built With Each)
✨ Episode Outline — click any point to jump to it in the episode
Problem solved
Most free-AI-tool lists only show logos; this episode shows what each of 10 free tools actually builds and how they connect into one private, local AI stack run by
Hermes Agent.
Benefits
- Private AI: Ollama counter shows zero cloud calls, zero bytes sent
- Hermes remembers past chats, saves solved tasks as reusable skills
- Obsidian vault acts as one shared memory for every agent
- Agent OS Kanban shows which agent did what, moved to done automatically
- Every tool on the list has a free way to start today
Use cases
- Perplexity research brief on how local physio clinics get patients online, sourced per claim, in under a minute
- Google AI Studio build mode wrote a physio clinic booking tool from one prompt in about five minutes
- Hallmark design skill produced three non-AI-looking landing page designs for the same physio clinic
- Claude skill 'Onboard Harbour Physio' writes welcome email, onboarding checklist and first-week plan end-to-end
- Gemma 4 SEO Lab audited a clinic page (title too long, no H1, no internal links) and returned the fixed version
KPIs / results
- Booking tool built in ~5 minutes from one prompt
- Hallmark ships ~50 real design themes plus an AI-tell checklist
- Ollama counter: cloud calls 0, data sent 0 bytes
- Gemma 4 E4B runs locally on a Mac; Apache 2.0 licensed
Tools / build
- Hermes Jarvis voice worker
- Gemma 4 SEO Lab (keywords, blogs, audit, tools)
- Agent OS workspace gallery + Kanban board on Hermes
- Nemotron vs Gemma side-by-side showcase page via Ollama
- Claude client-onboarding skill
I run my whole business on free AI. 10 tools. And for every single one, I'll show you the real thing I built with it. This is Jarvis. I talk and Hermes Agent does the job on my own computer. This is my Gemma 4 SEO Lab. Keywords, blog posts and a site audit from a free model. And this is my AgentOS workspace. Every build and every agent on one board. And every one of these 10 has a free way to start today. Most free tool lists just show you logos. This one shows you what each tool actually builds and how they all connect. So let's count down from 10 to 1. 10. Perplexity. Perplexity is an answer engine. You ask a question, it searches the web and every line comes back with its sources. Here's what I use it for. Before I build anything for a client, I need a research brief. So I asked Perplexity for one. Here's the prompt. Write a research brief on how local physio clinics get new patients online. Focus on local SEO, Google business profile, reviews, content and referrals. Cite every source. And in under a minute, I've got a brief. The main channels, what works and a numbered source next to every claim. I click a source and it opens the page it came from. That matters because AI can make things up. With sources, I can check every line before it goes near a client. You just go to perplexity.ai and start asking. The free plan gives you unlimited quick searches and a few deeper pro searches every day. Number nine, Google AI Studio. AI Studio is Google's playground for its models. And it has a build mode where you describe an app and it writes the whole thing. So I gave it one prompt. Build a booking tool for a physio clinic. Pick a service, pick a time slot, enter your details and show a confirmation. It wrote the app and it runs right there in the browser. I pick a sports massage. I pick Thursday the 12th at 10. I type a name and it books it. It even gave me a booking reference and the whole thing took about five minutes to build. That's a custom tool for a small business from one prompt with no developer. You sign in at AI Studio with a Google account, click build and type what you want. Number eight, Hallmark. Here's the problem with AI made pages. They all look the same. Same purple gradient, same cards, same font. Hallmark is an open source design skill from Together AI that fixes that. You add it to your coding agent and it brings about 50 real design themes, plus a checklist of the tells that make a page look AI made. So I took one landing page for the same physio clinic and asked for it in three premium designs. Here's the first one. Warm editorial, big serif headlines. Here's the second. Clean Swiss layout, strong grid. And the third, soft and calm with a lot of space. Same content, three looks. None of them look AI made. You install it with one command and from then on your agent designs with it every time it builds a page. Now, quick one. If you want my best free AI setups like the Jarvis Voiceworker, the Gemma 4 SEO Lab and the AgentOS Workspace, they're inside the AI Profit Boardroom. You get the builds from this video set up for you with a short guide for each. You get the skills and prompts I use every day, plus the full course and a community of business owners using this stuff. The link is in the comments and the description or go to AIprofitboardroom.com. Number seven, Obsidian. Obsidian is a notes app. Your notes are plain text files in a folder on your computer called a vault. And that's why it matters for AI. My agents can read the same folder I write in. So this is my vault. My offers, my SOPs, my brand voice, my client notes, all linked together in this graph. When Hermes writes a client email, it reads the brand voice note first. When Jarvis answers a question about a client, it reads the client note. And when I update a process, I change one note and every agent follows the new version the next time it runs. One memory for every note and every process. And I own it because it's just files on my computer. Number six, Nemetron. Nemetron is Nvidia's family of open models. I run Nemetron 3 Super through Olama, and I use it as a second brain. When an answer really matters, I don't want one model's opinion. So I built this showcase page in my workspace. I ask one question, and Nemetron and Gemma answer side by side. Here's the question. Write three headlines for a physio clinic's new back pain service. Gemma gives me three. Nemetron gives me three different ones. Now I pick the best, or I mix them. I do the same for anything important. A sales page, a client email, a plan. Two brains side by side. Another free brain, so you can compare answers before you use them. Number five, Claude with skills. Claude's free plan is a great chat assistant. But the real power is skills. And skills work on the free plan too. A skill is a folder of instructions that teaches Claude how to do one job your way. Every time. For anyone on your team. So here's one of mine running end-to-end in Claude code. A client onboarding skill. I type one line. Onboard Harbour Physio. And the skill takes over. It writes the welcome email, it builds the onboarding checklist, and it drafts the first week's plan. Same steps, same quality, every single time. That's a workflow anyone on the team can run, even if they've never written a prompt. You turn on code execution and skills in Claude settings, and you can upload your own skills as a zip file. Number four, Olama. Olama runs AI models on your own computer. You install it, you pull a model, and you're chatting. This is the Mac I'm recording on. Here's Gemma 4 running. Here's Quen. Here's the Jarvis voice brain. And here's the best part. Look at this counter. Cloud calls, zero. Data sent, zero bytes. Every answer is coming from this Mac. That's private AI. Client data, contracts, notes, nothing leaves the building. And Olama is what powers half of this list. Gemma, Nematron, and Jarvis all run through it. When a model is too big for my Mac, like Nematron, Olama runs it in the cloud with the exact same command. Okay, the top three. Number three, Agent OS. When you run a lot of AI, the hard part is keeping track of it. Which agent is doing what? Which build lives where? So I built Agent OS on top of Hermes, using its free dashboard and Kanban board. One dashboard for the whole AI team. The first screen is the workspace gallery. Every build from this video is here as a live card. The SEO lab, the booking tool, the Nematron page, the three hallmark designs. I click one and it opens. And the second screen is the Kanban board. Every task my agents are working on. To do, doing, done. Watch this one. Hermes picks up a task, moves it to doing, finishes it, and drops it in done. I didn't touch it. And every card shows which agent has it and what it produced. So at the end of the day, I can see exactly what got done. That's one place to run the whole AI team. Number two, Gemma 4. Gemma 4 is Google's newest open model family. It's free to download. It's open source under Apache 2.0. And the small versions run on a normal laptop. I run the E4B version on this Mac through Olama. And with it, I built the Gemma 4 SEO lab. A full SEO team from one free model. It has four tools. First, keywords. I type Physio Clinic Bristol and it maps out the keywords grouped by what the searcher wants. Second, blogs. I pick a keyword and it writes a full blog post with headings, a summary, and questions at the end. Third, the audit. I paste a page and it scores the title, the headings, the content, and the internal links with a fixed list. And fourth, the tools. Meta descriptions, schema, and title ideas in one click each. So here's a real run. I audit a clinic page. The title is too long, there's no main heading, and there are no internal links. And it hands me the fixed version. All of that running on my own Mac from a free model. One command to get it. Olama run Gemma 4. And number one. Hermes Agent. Hermes Agent is an open source AI agent from Noose Research. It runs on your own computer 24 hours a day. It has memory, skills, a schedule, and it can use any model you plug in, including the free ones on this list. It's the worker. Everything else on this list is something Hermes can use. Here's what makes it different from a chat window. It remembers. It searches its own past conversations, and it builds up a picture of you and your business over time. It learns. When it solves a new kind of task, it can save the steps as a skill, and it gets better at that skill the more it uses it. It works on a schedule. It has a built-in scheduler, so I give it jobs, like a morning report, and it sends the result wherever I want. And you can reach it from anywhere. Telegram, Discord, Slack, WhatsApp, or right here on the computer. And my favorite way to use it is Jarvis. I just talk. Jarvis, check the Kanban board and write the follow-up email for Harbor Physio. And Hermes does it. It reads the board, it reads the client note in Obsidian, it writes the email, and it moves the task to done. Then it tells me, out loud. Here's the email it wrote. It knows the clinic's name. It knows what we talked about last week. And it's in my voice, because it read my brand voice note. And while I'm away from the desk, the same Hermes keeps working through the board. That's a 24-7 AI worker on my own computer that I can just talk to. You install Hermes with one command from the Noose Research GitHub, pick a model, and it's running. So here's the whole stack. Hermes is the worker. Gemma, Nematron, and Olama are the brains. Obsidian is the memory. AgentOS is the dashboard. And Claude, Hallmark, AI Studio, and Perplexity are the specialists. If you're starting today, start with three. Install Alarma and pull Gemma 4. Install Hermes and point it at Gemma. Then make an Obsidian Vault with one note about your business. That's a private AI worker that knows your business in an afternoon. 10 tools, 10 real builds, all free. If you want all of this set up for you, here's what you get inside the AI Profit boardroom. You get Hermes Jarvis. You get the Gemma 4 SEO Lab. You get the AgentOS workspace all set up for you with a step-by-step guide for each one. You get the roadmap so you know exactly what to set up first. You get four live calls every week where you can ask me anything. You get new tutorials every single day. And if you get stuck, you can DM me. And here's what members are saying. The link is in the comments and the description. Or go to AI Profit Boardroom dot com. I'll see you inside.