Steve Peak — Own Your AI Agent

You Need Your Own Agent

Most people are renting AI by the API call, and that works right up until the bill doesn't, or the provider changes the rules, or your data's sitting on someone else's server doing someone else's math. I got tired of that math. So I built an agent that lives on your machine, it has memory, it has a whole toolbelt, and it talks to whatever it needs to talk to, including my own CRM (SeedlingCRM) if that's your flavor of chaos.

Agents like this are where everything is headed. Anyone still renting their intelligence by the token in five years is going to feel it. Might as well own yours now, and on Pro and Studio, you get me as your personal engineer: real human help getting set up, not a docs link (well you get that, too).

* Starter is self-serve with full documentation. Pro includes 1 hour of setup support. Studio adds 1 hour of monthly engineer support for any project you're working on.

Power

Full toolbelt out of the box: CRM access, email (read/draft/send, gated for safety), web research, image generation, long-term memory, and growing workflows for sales, marketing, and daily ops. Not a chatbot, a coworker.

Cost

Run it fully local and it's free to operate, no per-token bill, ever. Point it at Claude instead when you want more horsepower, billed to your own API key, your own budget, your own call.

Your Data

Everything it remembers, drafts, and generates lives on your machine. Nothing shipped off to train someone else's model. You see every tool it has, and every action it's about to take before it takes it.

Your Personal Engineer

Pro and Studio buyers get me on the line, not a ticket queue. I help you install, tune your models, wire up Seedling, and on Studio, an hour every month for any project: app ideas, automation, weird hardware, whatever you're building.

Claude, or fully local, your call

Claude backend

Smarter reasoning for harder problems, and no GPU needed at all. Billed to your own Anthropic key, you control the spend limit, and nothing runs through me.

Local backend

Runs on Ollama, entirely on your machine. Free, private, and switchable live from the dashboard, flip between the two whenever a task calls for it.

Not a chatbot. A coworker, with an engineer on call.

Your own expert, included on the right plans

I built this agent because I wanted one myself. When you buy Pro or Studio, you're not just getting software, you're getting a personal engineer in your corner. Setup calls, architecture questions, "can my agent do this?", that's the job. Studio goes further: an hour of engineer time every month, usable on the agent or anything else you're building.

Starter

Self-serve Full docs & wizard. You run the show.

Pro

1 hr setup support I get your agent running with you, live.

Studio

1 hr setup + 1 hr/mo Ongoing engineer time for the agent and your other projects.

A note on hardware

This is a full agent, it does real work, not just chat. Running it locally is free and private, but speed depends on your machine. No good GPU? No problem, the Claude backend runs in the cloud for pennies per use, so hardware stops mattering entirely. The setup wizard checks your machine and recommends the right choice automatically.

Tier	RAM	GPU	What you get
Minimum	8GB	Any / none	Use the Claude backend, no local GPU needed at all. Local mode works too, just with a smaller/slower model.
Recommended	16GB+	8–12GB VRAM (e.g. RTX 3060/4060/5070)	Comfortable local 14B-class models, responsive dashboard, image generation works well.
High-end	32GB+	16GB+ VRAM (RTX 4080/5080/5090, or 32GB+ unified memory on Apple Silicon)	Bigger local models, image generation and local chat running at once without competing for memory.

Not sure which you have? Any Windows/Mac machine from the last few years handles the Claude backend fine, the table only matters if you want to run models locally for free.

Get Your Own Agent

How Buying One Works

From "add to cart" to "my agent just drafted my emails" in an afternoon.

Pick a plan

One-time or monthly, whatever fits. Add plugins like image generation now, or bolt them on later.

License & download, instantly

Checkout issues your license key on the spot. Download from your account any time a new version ships.

Run the setup wizard

It checks your hardware, picks the right AI model for your machine, and installs everything. No manual config.

I'll help if you're stuck

Pro includes 1 hour of setup support; Studio adds 1 hour of monthly engineer support for any project. Not a docs link, me, on a call, getting it done with you.

See Plans & Pricing

Who's Behind This

I'm Steve, an all-purpose engineer. I grew up in the Silicon Valley taking things apart, and I've spent decades leading and building across software, fabrication, permaculture, and art. My skillset is finding paths others haven't: solving problems that haven't been solved, making things that haven't been made.

That's why this isn't a faceless SaaS product. When you buy an agent, you're getting software built with care by one person who answers his own support requests, and on Pro and Studio, acts as your personal expert engineer for setup and ongoing work. Backed by Seedling Systems LLC: no ads, no data selling, no dark patterns. Got a weird idea beyond the agent? I also take on projects, apps, products, interactive art for Burning Man, things without a job title.

Tell me what you want to build

Own your AI. I'll set you up.