A ladder of independent, vendor-neutral engagements — from first look to continuous assurance.
AI speed amplifies hidden weaknesses. Code that would have taken a team a quarter to write now lands in a week — and so do its flaws. Hallucinated APIs, shallow tests, over-permissioned agents, structural debt, and silent security regressions all ship at the same velocity as the features that look impressive in a demo. The faster a codebase grows under AI assistance, the more it needs an independent party whose only job is to see what the speed conceals.
The Guild offers that party. Our services are arranged as a ladder, because the right level of scrutiny depends on where you are. You can step on at any rung. You can climb as far as your system warrants. We sell no tooling, so every recommendation is the one we would give a friend.
A single-page overview of the full Guild ladder — Diagnostic, Audit, Optimization, and Continuous Security Assurance — with what each rung covers, when to step on, and how the rungs connect. Useful for circulating internally when deciding which engagement fits your situation.
Before the ladder, a signal. A three-minute, private self-assessment for engineering leaders — no code access required.
Ten questions about how your team uses AI, how your codebase feels to work in, whether your tests and dashboards can be trusted, what your agents are permitted to do, where your token spend is going, and the one quiet worry you'd want an independent second opinion on. Six of the answers feed a provisional signal band; four are open-ended and speak in your own words.
How your answers become recommended next steps. When you submit, the Guild receives the full picture — your rated answers, the flagged soft spots, the provisional band (1 through 4), and the free-text passages describing what has gotten harder, what people quietly avoid changing, and the hunch you can't yet prove. That is enough for a human at the Guild to write you a short, candid read: whether the signals warrant a deeper look, which specific patterns are pointing in the same direction, and which rung of the ladder below is the honest starting point for your situation. If your answers say you're in good shape, that is exactly what the read will tell you. There is no pitch, no funnel, and no upsell — the questionnaire exists so that any recommendation we make is grounded in what you actually shared, not in what we happen to sell.
Four rungs, each building on the last. Start where you need to start; climb as far as your stakes require.
A short, focused conversation and surface assessment of your AI-assisted development practice. We look at how AI is being used, where the obvious risks sit, and whether deeper review is warranted. The diagnostic is the cheapest way to find out whether you have a problem worth solving — and it ends with an honest answer, not a sales pitch.
Independent, candid review of an AI-assisted codebase: hallucination detection, security vulnerabilities, test integrity, dependency and license risk, architecture coherence, data handling, error handling, and agent configuration. The audit tells you what your codebase actually contains — not what the tooling reports it contains.
Once the audit has surfaced what is broken, bloated, or wasteful, optimization removes it. We reduce token waste, eliminate dead weight, tighten structure, and lower the ongoing cost of running and maintaining an AI-assisted codebase. The audit finds the problems; optimization eliminates the conditions that create them.
A standing, vendor-neutral relationship that verifies your AI-assisted code and agent behaviour continue to meet the open security canon — OWASP Agentic Top 10, OWASP LLM Top 10, SAFE-MCP, NIST AI Agent Standards, and applicable regulatory regimes — on the cadence your work actually moves at. Point-in-time audits expire; continuous assurance does not.
The market is filling with instruments. Security fabrics, governance platforms, evaluation harnesses, agent monitors — each one a watcher, each one sold by a vendor with a roadmap to protect and a story to tell about its own coverage. Useful tools. But a tool cannot grade itself, and a vendor cannot grade its competitors.
The Guild's ambition is to become the impartial layer that watches the watchers.
We are the party with nothing to sell except the accuracy of our assessment. We verify whether your tooling is configured to enforce the standards it claims to enforce. We check whether your audits cover the surface that actually matters. We tell you when a dashboard is reassuring and wrong. And because the standards evolve and your system evolves, we do it on a rhythm that keeps up with both.
Independence is not a marketing posture. It is a structural commitment: no platform, no fabric, no proprietary taxonomy, no referral fees. Just craft-grounded human judgment, applied honestly, against open standards the whole industry can read.
The ladder is sequential by design, but you do not have to start at the bottom.
Start with the Diagnostic. It is the lowest-cost, lowest-commitment way to find out whether deeper review is warranted.
Start with an Audit. Independent scrutiny before launch is far cheaper than after.
Move to Optimization. The findings are only valuable once they are remediated.
Engage Continuous Security Assurance. Point-in-time review cannot keep up with agentic systems.
Independent, candid review of AI-assisted codebases for correctness, security, and structural integrity.
Reducing waste, improving efficiency, and lowering the token cost of AI-assisted software.
Vendor-neutral, continuous verification that your AI-assisted code and agent behaviour meet open security standards.
Return to the AI Craftspeople Guild home page.