Web3 in Travel

The first build-in-public episode of Web3 in Travel. Instead of telling you what I've learned about AI, I build an AI travel agent on screen, unscripted, with Claude Code and Codex, and you watch every step, including the parts that break.There are two agents. The first works for the traveller: it gets to know you, then finds where to stay. The second will work for the hotel: it reads the emails of guests who want to book direct, and later it could run on a local AI so guest data never leaves the hotel. The code will be open source, and the agents write the blog about it.In this episode I test the traveller agent with Lollo, a made-up business traveller, search for three nights in Munich, and check what each search costs. Then I compare the results with Booking.com by hand, and hit the real blocker: the data. The AI was picking the best from a bad list. We fix it live with filters (Google 4.5+, at most €220 a night) and ranking by distance, and find hotels as good as Booking's, one of them €46 a night cheaper.Collaborations: [email protected]:00 From telling to building7:43 Where we left off11:51 My setup22:36 Meet Lollo29:11 Built in a meeting30:54 Searching for hotels37:28 What a search costs53:18 Booking.com by hand62:43 The data problem66:58 The fix

We are in the middle of a bug apocalypse. With AI, finding serious software bugs has become fast and cheap, for attackers and defenders alike. In October a critical PHP bug made governments, ports and hospitals take their own websites offline while they patched.I'm not a developer, yet I found a bug in the Sparrow Bitcoin wallet that the experts had missed. What that says about who builds now, how to attack your own code with AI before someone else does, how I cut my blog's attack surface by moving from WordPress to plain HTML, and why DeFi, the best honeypot for hackers, may come out of this safer than closed legacy software.Collaborations: [email protected]:00 Intro0:20 The bug apocalypse0:51 Both sides got new weapons1:55 The PHP bug3:13 What it means for us3:34 The bottleneck moved5:11 My bug in a Bitcoin wallet6:06 From how to what8:26 Your software is in the war9:24 Attack your own code with AI10:38 Software without bugs is impossible11:35 Fewer doors: the attack surface12:45 My blog: from WordPress to pure HTML15:55 One door left17:22 When non-developers send fixes18:54 Crypto: the best honeypot20:33 Closed software isn't safe either20:54 Immutable contracts21:19 Web3 in Travel's end game?21:39 What to do now

Can a laptop do your company's AI work without sending a single byte outside? Luca De Giglio ran real travel-business tasks on a MacBook Pro with 48 GB of RAM, using the open-weight Qwen 3.8 27B model, and compared every answer with Claude Opus 5.5 through the API.On everyday tasks (reading a booking email, replying to a bad review, pricing a tour package) the local model matched Claude, at about a fifth of the speed and a fraction of the cost, and the data never left the laptop. On a 30-page supplier contract it still found the right cancellation charge, but in six minutes instead of sixteen seconds. On 600 bookings checked against a hotel invoice, it hit a wall.Then the questions that matter for a company: why more RAM is not enough, what actually fits on a laptop, why buying a machine is a bet, who maintains it, and where local AI makes sense in travel and hospitality.IN THIS EPISODE• Three light tests and two heavy tests: local vs. the Claude API.• Why a 64 GB server can be useless for AI, and a 48 GB Mac isn't.• The model as a book: capacity and bandwidth.• Ollama, LM Studio, and models tuned for one machine (DwarfStar, Sushi).• The money: 50× cheaper on electricity, and why that's rarely the point.• Maintenance, evals, and why private isn't automatically safe.• Guest data, contracts and margins: the travel use cases.CHAPTERS0:00 Intro0:20 Local AI: is it time to start?1:43 My laptop: 48 GB of RAM3:16 Why a big server is useless for AI4:08 Test 1: a booking email5:43 Test 2: reply to a bad review7:54 Test 3: a tour package margin9:22 Speed: tokens per second10:12 Cost, and where the data went11:25 Heavy test A: a 30-page contract14:05 Heavy test B: 600 bookings vs. an invoice16:43 Electricity: 50× cheaper17:25 The model is a book19:01 What runs on 48 GB, and the tools19:56 Models tuned for one machine22:29 Does local save you money?23:57 Buying a machine is a bet26:09 Finding the ceiling, and evals28:07 Maintenance28:53 Private isn't automatically safe29:32 Local AI in travel and hospitality30:57 A possible setup, and where to startFor collaborations, advisory work or speaking:[email protected]://www.linkedin.com/in/luca-de-giglio/

Most companies are still asking how to use AI. The next question is how to use it privately. When your team plugs Claude or ChatGPT into the company's documents, where does that data go, and what can the AI labs do with it?In this episode of Web3 in Travel, Luca De Giglio explains why your data is worth far more to an AI lab than it ever was to Google, why a $20 subscription buys hundreds of dollars of inference, and what "we don't train on your data" really promises. He then walks through the options for private AI, from the most private to the least: local open-weight models on your own hardware, private APIs for open models, and Trusted Execution Environments (TEEs). For each, what it costs and what it actually protects.He closes with the strategy: route each task to the right model, decide which data must never leave the company, and accept that total privacy does not exist. You manage risk, you don't cancel it.IN THIS EPISODE• Why LLMs extract more value from your data than ads ever did.• The real economics of a $20 AI subscription.• Could an AI lab become a competitor in your industry?• Local models and open weights: DeepSeek, Kimi K3, Qwen, gpt-oss.• Private APIs and anonymized inference: what they cost, how private they are.• TEEs: hardware attestation, and why you still have to verify.• Routing: the right task to the right model.• The first step: decide which data never goes near an LLM.MENTIONED• Vitalik Buterin on a local model deciding what to send to frontier models: https://x.com/VitalikButerin/status/2106537633056969024CHAPTERS0:00 Intro0:20 The next question: how do I use AI privately?0:59 Google sold ads. LLMs take the value1:33 Your best insight, in the next model2:39 Why a $20 subscription is so cheap4:34 Do they train on your data?5:24 When the AI lab enters your industry7:51 Local models and open weights10:07 The cost of running AI yourself11:25 Private APIs: paying full price13:08 How private is anonymized?14:21 TEEs: trusted execution environments16:12 Routing: the right task to the right model17:43 A strategy for your data19:48 The first step: data that never leaves21:21 Nothing is 100% privateFor collaborations, advisory work or speaking:[email protected]://www.linkedin.com/in/luca-de-giglio/

The real leverage from AI starts when you go beyond ChatGPT: a project with agents for each area of your work, in tools like Claude Code or Codex that build their own tools.I share three people I onboarded and what changed for them, why these coding tools are for non-coders too, and why the terminal stops most people. I introduce The White, the software I use to onboard people, and explain who succeeds, who fails, and why you should start small.Want help making the jump? [email protected]://www.linkedin.com/in/luca-de-giglioCHAPTERS0:00 Intro0:20 Why I share new things early1:35 Beyond ChatGPT: a project for everything3:35 How do you make the jump?4:29 Three people I onboarded5:30 Coding tools are not just for coders6:38 A third way to get software7:55 How one CEO automated his company10:45 "What did you do to him?"13:57 The terminal problem and The White16:18 An offer for listeners17:24 Who succeeds19:20 Why AI companies want you out of the chat21:23 Who fails23:03 Start small24:29 Not a CEO? Become a secret superhero

I could not find how to cancel limit orders in my Safe wallet. Claude Code built a JSON transaction file; I imported it, ran a simulation and signed the cancellation myself.I also share my loss-making Token Trade experiment, why I prefer chat to complicated dashboards, and the possibilities for agent payments. The closing discussion explores AI wallet risks, decentralized travel and Trips Community, and a future where we express our intent to an agent. I am not claiming AI wallets are safer or right for every use case.Collaborations: [email protected]:00 Intro0:20 Wallets and the adoption hurdle3:40 Convenience, custody and attention5:26 The Safe limit-order problem7:54 JSON, simulation and signatures11:23 Building Token Trade14:15 Trading losses and on-chain exchanges19:12 Building tools through chat20:42 Why dashboards drift23:13 Stablecoins or credit cards?25:40 AI wallets and security trade-offs26:50 What agents could change in travel28:35 Intent as the interface

The step most people have not taken with AI: from chatting with ChatGPT, a very smart consultant you phone and re-brief every time, to Claude Code working inside your own files, like someone in your office who remembers everything.I explain why that step looks like a mountain and is really a bump, then show my actual setup: Claude Code plus one folder, moved onto a server that works 24/7, one container per project, about forty of them, laid out as a pixelated city. Taxes, property, a China trip planner, this podcast's own pipeline, and a local AI whose job is to keep me focused.Collaborations: [email protected]:00 Intro0:20 What I Actually Do4:00 ChatGPT Is a Consultant on the Phone6:00 The Bump That Looks Like a Mountain9:00 Someone in Your Office Who Builds Tools15:40 Claude Code and One Folder22:20 Moving It to a Server That Works 24/726:00 One Container per Project, and the City32:30 The Business District37:20 Travel, and Why Chat Beats Dashboards42:10 The Coordinator and a Local Consigliere46:40 New Projects in a Minute, and the Stack

An AI agent needed data from a website, so it found Apify, set up its own crypto wallet in thirty seconds, and paid one cent through x402 — no account, no credit card, no API key, no human.I unpack Erik Voorhees' tweet about that transaction, explain how x402 turns the dormant HTTP 402 "Payment Required" code into a payment rail for machines, and why stablecoins cover the sub-cent payments credit cards cannot.Then the travel angle: Sabre, PayPal and Mindtrip's agentic booking system — settled in stablecoins, not cards.Collaborations: [email protected]:00 Intro0:20 Voorhees' Agentic Payment Experiment3:40 Apify Purchase via x4027:20 Gasless USDC Explained9:40 How x402 Payments Work12:40 Why Micropayments Need Stablecoins16:00 Pay-Per-Use Agent Commerce18:20 Agent Wallets and Identity21:40 Stablecoin Travel Bookings

Using AI, Luca De Giglio found an exotic bug in a Bitcoin wallet despite not being a Bitcoin developer—and the project’s developers accepted and merged the fix. That experience anchors the return of Web3 in Travel for a second season focused on what happens when AI becomes a practical building tool rather than a chat interface. Luca explains how he manages 40 active projects, runs most of his work on remote computers and experiments with local models, coding agents and multi-model verification. He also considers what these tools mean for development teams that either reject AI entirely or trust it without adequate checks. The discussion then connects AI with blockchain infrastructure: agents can create wallets, obtain funding and autonomously pay for API access using USDC and X402. For travel and technology leaders, the larger question is no longer simply whether AI can complete a task, but which consequential problems are now worth attempting—and how organizations should adapt their workflows, security practices and ambitions.CHAPTERS0:00 Intro0:20 Podcast Returns With AI4:00 Forty Projects and Context Switching6:40 Remote AI Coding Workflow8:20 Local Models and AI Learning12:00 AI Finds Crypto Wallet Bugs17:40 Developers Resisting AI19:40 Stripe Embraces AI and Blockchain22:20 Agents Paying Through x402IN THIS EPISODE• Why the podcast is returning with AI alongside its original blockchain and travel focus• The challenge of managing 40 projects while coding agents create constant context switching• How remote development keeps AI tasks running after the laptop is closed• Using Claude Code, Codex and Kimi to cross-check a suspected crypto software bug• Why both rejecting AI-generated code and trusting it uncritically create risks• How local models change the economics and pace of AI-assisted development• An example of an agent using a wallet, USDC and X402 to purchase API access—I don't understand things unless I do them, so I'm constantly building with blockchain and AI — over 40 projects at once. This podcast is where I share the most interesting learnings.For collaborations, advisory work or speaking:[email protected]://www.linkedin.com/in/luca-de-giglio/

I just got back from three days at ITB Berlin, walking the halls, talking to founders, payment companies, and travel executives.AI was the topic everywhere. Everyone knows it matters. But what struck me is how early we still are. Most people have experimented with tools like ChatGPT, yet very few teams are actually using AI inside their real workflows.The big shift seems to happen when people start working with tools like Claude Code or OpenAI Codex and point them at their own folders, data, and processes. Suddenly AI stops being a chatbot and becomes something you can actually delegate work to.In this episode I share a few things that stood out from ITB:– why many travel companies feel pressure around AI but don’t know where to start– why the next wave of innovation might come from very small AI tools solving niche operational problems– why stablecoins and crypto payments are still barely discussed in the industry– and an experiment I’m starting with a public AI agent that tries to negotiate hotel pricesThink of it as a quick field report from inside the travel tech ecosystem.