The 48-hour AI trial: what you can actually build from just your website URL

Give Brainztem your website URL and it builds you a working preview of your own agentic operating system: your name, brand, and voice pulled from your site; a credibility score of the site itself; a crew of AI agents staffed for your kind of business; and a brain seeded with everything your site says about what you do. The trial is free, lasts 48 hours, and includes ten real conversations with your agents. No credit card, no sales call, no uploads — the URL is the entire form.

This post is a straight walkthrough of what happens in those 48 hours: what the scan actually builds, what the honest limits of a trial instance are, and how to use your window well.

Why a URL is enough

Your website is the densest public artifact your business has. It states what you sell, to whom, in what tone, from where. It carries your colors, your logo, your services, your proof. Most software trials ignore all of that and hand you an empty workspace with sample data — then ask you to imagine. We think the imagining is the product's job.

There's a quieter reason too: a URL is information you've already published. The trial asks for nothing private — no inbox access, no document uploads, no contact list. You get to judge the crew's usefulness at zero incremental exposure, and the system has to earn anything more. We wrote about that staging principle in our guide to sharing context without leaking data.

What the scan builds, piece by piece

Head to brainztem.com/start, enter your URL, and watch the build happen live rather than behind a spinner:

• Your identity. The scan extracts your business name, tagline, summary, audience, and region — plus your brand accent color and logo, so the workspace that appears is unmistakably yours, not a demo wearing your URL.
• Your credibility score. Your site is scored out of 100 with the engine from our sister product, websitecreditscore.com — an instant read on how credible your web presence looks to a stranger, with the dimensions behind the number. This alone is worth the visit even if you close the tab after.
• Your crew. Agents are hired to fit the business the scan found — the roles a grants consultancy needs aren't the ones a fabrication shop needs — each introduced with a name and a job.
• Your seeded brain. What your site says about services, positioning, and audience becomes the starting knowledge graph your agents reason from, so their first answers are about your business, not a generic one.

The result isn't a slideshow about what Brainztem would be like. It's a small, live instance of Brainztem being yours. That distinction matters for evaluation: a demo shows you the vendor's best case, assembled from a fictional company chosen to flatter the product. A preview built from your own site shows you the awkward, specific reality — how the system handles your niche, your regional audience, your slightly odd service mix — which is the only version of the product you'd actually be buying.

What you can do inside the window

The trial gives you a sandbox workspace and ten chat messages with your agents — enough for a real evaluation if you spend them deliberately. Talk to the crew about your actual business: ask what they'd prioritize from your pipeline, have one draft an outreach email in your voice, ask another how it would structure a client follow-up sequence. Because the brain is seeded from your site, the answers name your services and speak in your register — which is precisely what you're there to judge.

You can also work the workspace itself: explore your score, review the crew, add tasks and notes. Everything you do persists for the full 48 hours, and here's the part that rewards effort — if you convert, your sandbox carries over into your real instance. The trial isn't a throwaway demo; it's the first 48 hours of the actual system. Drop your email inside the trial and the workspace survives the window with you.

Most trials show you the software. This one shows you your business, already running on it.

The honest limits

A trial built from public data has ceilings, and you should know them going in. The brain only knows what your website says — if your site undersells you, your preview will too (the credibility score will usually have told you that already). The ten-message budget means you should ask real questions, not warm-up ones; paid instances raise the ceiling to fifty per sandbox and full access beyond it. Trial agents prepare and draft, but connected actions — actually sending email, touching live accounts — belong to full instances, where they run behind human approval gates anyway. And at 48 hours, an unclaimed trial expires and unpaid sandboxes are purged. The countdown is real; we built it that way on purpose.

How to spend your 48 hours well

A simple plan: first hour, run the scan and read your credibility score — note what it flags, because that's free consulting. First day, spend five of your ten messages on your hardest operational headache and judge whether the answers show real understanding of your business. Second day, spend the rest pressure-testing the thing you'd delegate first — outreach drafts, follow-up structure, pipeline triage — and show a draft to someone whose taste you trust. If the crew's work would survive your review queue, you have your answer. From there, milestone one ($3,000 of the $10,500 engagement) unlocks full access, and everything you built in the sandbox comes with you.

The whole thing starts with the one piece of data you've already published. Enter your URL and watch your instance build itself — the first 48 hours are on us.