Frequently asked questions

Lobster Pack sets up AI agents for European businesses. You tell me what's eating your time, and I deploy the right tool — OpenClaw, Claude Code, or something custom — on your hardware. I handle configuration, integrations, security hardening, and training. It's a one-person operation backed by 15 years of engineering experience, including a decade at MIT Media Lab.

OpenClaw is an open-source AI assistant released in November 2025. It can handle real tasks: managing email, updating your CRM, drafting documents, navigating websites, scheduling meetings, and more. It hit 248,000 GitHub stars within months. The reason it matters is that it's free, it's powerful, and it can run fully on your own hardware — meaning your data doesn't have to touch anyone else's servers if you don't want it to. That last part is a big deal for businesses that care about privacy.

ChatGPT is a chatbot — you type a question, it types an answer. OpenClaw is an agent. It can actually do things: open your browser, fill out forms, send emails, move files around, update spreadsheets. Think of ChatGPT as a person you can talk to, and OpenClaw as a person who can also use your computer. The other big difference: OpenClaw can run on your own machines if privacy matters to you. With ChatGPT, everything always goes through OpenAI's servers — there's no local option.

No. I deploy OpenClaw, Claude Code, and custom setups — whatever fits. OpenClaw is great for general business automation, Claude Code shines for development and technical workflows, and sometimes a custom build makes more sense. The AI landscape moves fast. I'm not locked into one vendor, and if a better tool comes out next week, I'll tell you about it.

Here are some real examples: sort and respond to routine emails, update your CRM after every client call, generate invoices and reports, schedule meetings across time zones, draft proposals from templates, monitor competitors' websites, and process incoming applications or orders. The common thread is repetitive work that follows patterns. If you or your team spend hours doing something predictable, there's a good chance an agent can handle it.

Depends on which setup you choose. If privacy is a priority, I deploy a local model on a Mac Mini in your office — nothing leaves your building. Zero data goes to OpenAI, Google, or Anthropic. If you'd rather save money or get access to stronger models, I can wire things up through a cloud AI provider instead. Your data would pass through their servers, but that's a tradeoff some businesses are comfortable with. There's also a middle option: a dedicated GPU server running an open-source model, hosted by a third party but controlled by us. We'll talk through it during the free call so you pick what actually fits.

Yes, with proper configuration. Out of the box, OpenClaw is designed for single-user setups. For business use, I harden the deployment: access controls, network isolation, logging, permission boundaries so the agent can only touch what it's supposed to. I also set up monitoring so you can see exactly what the agent is doing at any time. It's open-source software with a huge community reviewing the code, which is generally safer than closed-source alternatives.

I take GDPR seriously — I'm based in the EU and subject to it myself. When I deploy AI tools for you, personal data stays on your infrastructure within the EU. I don't transfer client data to third countries. If the engagement involves processing personal data on your behalf, we'll sign a Data Processing Agreement. I can also help you with the data protection impact assessment (DPIA) that some AI deployments require under GDPR.

The EU AI Act is being phased in through 2026. Most business automation use cases — email sorting, CRM updates, document generation — fall into the minimal or limited risk category, which means lighter requirements. I stay current on the regulations and will flag anything that might affect your deployment. If your use case touches high-risk territory (like automated decision-making about employees or customers), we'll discuss compliance requirements before building anything.

A full deployment — where I install, configure, and integrate an AI agent into your workflow — starts at EUR 1,100 for remote, or EUR 2,800 plus travel for on-site. Monthly support retainers start at EUR 230/month. The discovery call is free. I'll always give you a clear quote before work begins. You can see the full breakdown on the pricing page.

No. The deployment is a one-time project. Once it's set up and working, it's yours. Many clients do choose a support retainer because the AI landscape changes fast and they want someone keeping their setup current. But it's not required. You can always come back later if you need help.

Absolutely. The discovery call is free — 30 minutes, no obligation. We'll talk about your business, where you're spending time on repetitive work, and whether AI automation makes sense. If it doesn't, I'll tell you that. You can book a call here.

Bank transfers (SEPA), credit cards, and PayPal. All invoices are in euros. For fixed-price projects, it's 50% upfront and 50% on completion. Retainers are invoiced monthly. Nothing complicated.

A Mac Mini (~EUR 700) works great for most small businesses — its unified memory lets both the CPU and GPU access the same RAM, which is ideal for running AI models locally. A desktop you already own is fine too, as long as it has at least 16 GB of RAM. If you'd rather not run hardware in your office, a VPS from EUR 20/month does the job. For bigger workloads, 32 GB of RAM and a dedicated GPU help. I'll help you figure out the right setup during the discovery call.

Almost certainly. The AI agents I deploy can interact with anything that has a web interface or an API. That includes Google Workspace, Microsoft 365, Slack, most CRMs (Salesforce, HubSpot, Pipedrive), accounting software, project management tools, and custom internal apps. During the discovery call, I'll map out which of your tools can be connected.

The initial setup — getting the agent installed and your first workflows running — typically takes 1 to 2 days. The full deployment depends on what you're automating and how many systems need connecting, but most projects wrap up within a week. More involved setups with many integrations, custom workflows, and team training can take 2 to 3 weeks. I'll give you a realistic timeline before we start. No one benefits from me underestimating the work.

If you're on a support retainer, you contact me directly and I fix it. Response time depends on your plan, but I aim for same-day responses for anything that blocks your work. If you're not on a retainer, you can still reach out — I'll help on an hourly basis. I also set up monitoring and alerts during deployment, so many issues get caught before you even notice them.