How AI works at Seglamater Services

AI is on our team. So is the cybersecurity professional running this company. The work you get reflects both — and we think you should know exactly how that division of labor plays out before you hire us.

The split

What AI handles

• Drafting emails, scopes, documentation
• Writing initial versions of code (always reviewed before ship)
• Scanning websites for privacy and security issues (the scanner itself runs deterministically — it's not an LLM guessing)
• Monitoring our own infrastructure and surfacing anomalies
• Researching vendors and technologies
• Summarizing log output, audit findings, technical documents

What a human handles

• Every email you receive — read and adjusted before it goes out
• Every recommendation we make to you — cross-checked against your actual environment, not what AI assumes about it
• Every line of code that ships to production — reviewed and merged manually
• Every customer conversation
• Every decision about which tool, service, or vendor to recommend
• Every security incident, from triage through remediation

The guardrails

The boring infrastructure that keeps AI from going off the rails:

• Code review on every commit. Nothing ships to production without an explicit human merge.
• Pre-commit security scans. Hooks block obvious mistakes — committed credentials, hardcoded secrets, sensitive file patterns.
• Cosign-signed releases. Anything we deliver to a customer is signed with a key only we hold. The signature is verifiable; we can share the public key on request.
• AI provider terms you can read. When we use Anthropic Claude for customer-facing work, we use it under Anthropic's Commercial Terms of Service, which state that inputs and outputs from the API are not used to train Anthropic's models. Anthropic's privacy center answers this directly. We don't opt into any training-data sharing programs.
• Audit trail in writing. Every significant change to our infrastructure or to a customer's environment is tracked, with a record of why it was made and what was tested. We can walk a prospective client through that record on request.

The proof

Things you can verify yourself, without taking our word for it:

• Our scanner runs deterministic checks, not AI guesses. The Seglamater Privacy Standard is a rule-based scanner that runs 31 specific checks against your site. The criteria are documented in our public specification — read it before you trust the grade. We can walk a prospective client through the source code on request.
• We grade our own sites. Scan us yourself: seglamater.app/privacy/scan/seglamater.com. Every Seglamater property scores A or A+ — if we're recommending you tighten your privacy posture, we've already done it on our own work first.
• We document what we do. Our workshop at seglamater.app is a running log of work in progress — what's broken, what's fixed, what we tried that didn't work. If we won't write about something publicly, we won't sell it to you either.

The honest limits

Where AI doesn't get to drive:

• Your specific business context. AI can summarize an industry, but it doesn't know the people who answer your phones, the regulators in your space, or the contract terms you signed last year. Every recommendation goes through a human who's actually talked to you.
• Legal questions. We don't use AI to interpret your contracts, your compliance obligations, or anyone's terms of service. That's lawyer work, not AI work.
• Security incidents. When something goes wrong — yours or ours — a human runs the response. AI is welcome to help research and draft, but it doesn't decide what to do.
• Vendor recommendations. AI can list options. Picking which one is right for you is judgment work, and it's mine.

What this means for the work you get

• You don't pay for AI overhead. AI does the work AI does well, which keeps our prices low without cutting corners.
• You don't pay for AI slop either. Every deliverable, every email, every recommendation has a human's eyes on it before it reaches you.
• Faster turnaround on routine work. Drafts, scheduling, scoping a quote — these don't take days because AI does the first pass.
• Same depth on the work that matters. When you ask a question that requires actual thinking, you're getting actual thinking, from someone who's been in cybersecurity professionally.

Who's behind this

I'm Quinn. I've been working with technology since I was under 10 — taking things apart, asking why, teaching myself Linux and security the way most kids learn sports. Over 20 years in IT later, I run Seglamater Services because the everyday person — the small business owner, the family, the person handed software they didn't choose — deserves the same quality of privacy and security help that big companies pay top dollar for.

The work here is education-first. The goal isn't to fix your problem this one time and bill you for it again next month. The goal is to make sure you understand it well enough that you don't need me next time. That's the bar I hold for every conversation, every recommendation, and every line of code that ships with my name on it.

I write longer about why I work this way at seglamater.app/about. Read that if you want the full picture before you hire me.