# DeployGuild

> DeployGuild is a forward-deployed AI engineering agency. It embeds FDEs to diagnose stalled AI pilots, ship production workflows, and leave evals, controls, runbooks, adoption proof, and handoff artifacts inside real operating environments.

## Core Pages

- [Home](https://deployguild.dev/): Overview of DeployGuild's FDE agency services, deployment loop, proof artifacts, AI rescue work, readiness scorer, and handoff discipline.
- [Book a Workflow Diagnostic](https://deployguild.dev/book-diagnostic): Buyer-facing diagnostic brief that forces one AI workflow into view before booking: current state, failure modes, production bar, artifacts, access, and handoff context.
- [Apply as an FDE](https://deployguild.dev/apply): Supply-side application page for builders who want to be reviewed for DeployGuild FDE work.
- [Field Notes](https://deployguild.dev/blog): DeployGuild essays about forward-deployed engineering, enterprise AI deployment, evals, workflow fit, and proof-led AI work.
- [AI Deployment Readiness Scorer](https://deployguild.dev/tools/ai-deployment-readiness): Public diagnostic tool that scores one AI pilot, RAG assistant, copilot, workflow automation, or agent against production-readiness evidence: evals, permissions, monitoring, security, rollback, adoption, and handoff.
- [Sample AI Deployment Handoff Dossier](https://deployguild.dev/sample-rescue-dossier): Public composite artifact showing what serious AI deployment work should leave behind: findings, decisions, scope, blockers, eval plan, rollback path, and handoff checklist.

## Primary Article

- [AI Deployment Checklist: 15 Things to Prove Before a Pilot Becomes Production](https://deployguild.dev/blog/ai-deployment-checklist-pilot-to-production): A practical checklist for moving AI pilots, RAG assistants, agents, copilots, and workflow automations into production with evals, security, monitoring, adoption, rollback, and handoff.
- [Markdown copy](https://deployguild.dev/content/ai-deployment-checklist-pilot-to-production.md): Plain Markdown version of the AI deployment checklist.
- [What Is a Forward Deployed Engineer? The AI Deployment Role Explained](https://deployguild.dev/blog/what-is-a-forward-deployed-engineer-ai-deployment-role): A practical definition of the FDE role, what FDEs do, how they differ from solutions engineers and consultants, and how engineers can prove the full deployment loop.
- [Markdown copy](https://deployguild.dev/content/what-is-a-forward-deployed-engineer-ai-deployment-role.md): Plain Markdown version of the FDE role guide.
- [Intelligence Does Not Deploy Itself](https://deployguild.dev/blog/intelligence-does-not-deploy-itself): Why OpenAI, Anthropic, Lovable, ServiceNow, Accenture, Scale AI, Palantir, and others are hiring Forward Deployed Engineers, and what the rush admits about enterprise AI.
- [Markdown copy](https://deployguild.dev/content/intelligence-does-not-deploy-itself.md): Plain Markdown version of the article for agents and crawlers that prefer text-first content.
- [Evals Are the Contract](https://deployguild.dev/blog/evals-are-the-contract-ai-deployment): How to build an eval set before you trust an AI deployment: collect real cases, include the failures, write expectations first, grade repeatably, and treat the eval set as the most transferable handoff artifact.
- [Markdown copy](https://deployguild.dev/content/evals-are-the-contract-ai-deployment.md): Plain Markdown version of the evals field note.
- [Prompt Injection Is a Deployment Problem, Not a Model Problem](https://deployguild.dev/blog/prompt-injection-is-a-deployment-problem): Why prompt injection is structural and cannot be patched out by a better model, and the deployment-layer defenses that contain it: least privilege, human approval gates, untrusted-by-default content, validated tool inputs, and logged traces.
- [Markdown copy](https://deployguild.dev/content/prompt-injection-is-a-deployment-problem.md): Plain Markdown version of the prompt injection field note.
- [Why Your RAG Assistant Is Confidently Wrong](https://deployguild.dev/blog/why-your-rag-assistant-is-confidently-wrong): Most RAG failures are retrieval, freshness, chunking, and permission failures, not model failures. How to diagnose confident wrong answers and fix them in the pipeline, with grounding as the non-negotiable.
- [Markdown copy](https://deployguild.dev/content/why-your-rag-assistant-is-confidently-wrong.md): Plain Markdown version of the RAG rescue field note.
- [Adoption Is the Last Mile](https://deployguild.dev/blog/adoption-is-the-last-mile): Why technically perfect AI systems still fail when nobody uses them, why users abandon working tools, and how to design for behavior change and measure adoption instead of launch.
- [Markdown copy](https://deployguild.dev/content/adoption-is-the-last-mile.md): Plain Markdown version of the adoption field note.
- [The Handoff Artifact](https://deployguild.dev/blog/the-handoff-artifact-ai-deployment): What a real AI deployment leaves behind so another engineer can own it: runbook, eval set, data and permission map, decision record, and adoption record, which also serve as redacted proof of confidential work.
- [Markdown copy](https://deployguild.dev/content/the-handoff-artifact-ai-deployment.md): Plain Markdown version of the handoff field note.
- [The Most Valuable Thing an FDE Says Is No](https://deployguild.dev/blog/the-most-valuable-thing-an-fde-says-is-no): Scoping is where deployments are won or lost. Discovery before scoping, decomposing the workflow into the right treatments, and the forms of saying no: not this part, not yet, not without a gate, not at all.
- [Markdown copy](https://deployguild.dev/content/the-most-valuable-thing-an-fde-says-is-no.md): Plain Markdown version of the scoping field note.
- [How to Prove Forward-Deployed Work When the Evidence Is Confidential](https://deployguild.dev/blog/proving-confidential-deployment-work): Building legible proof of confidential deployment work without breaching an NDA: redacted artifacts, a structured account of the deployment loop, clean-room reconstructions, references, and why the profession needs a shared standard.
- [Markdown copy](https://deployguild.dev/content/proving-confidential-deployment-work.md): Plain Markdown version of the proof field note.

## Key Concepts

- Forward Deployed Engineers are the deployment layer between AI capability and production business value.
- The hard bottlenecks in enterprise AI are workflow fit, data fit, eval fit, security fit, and adoption fit.
- AI pilots should not be called production until they have owners, evals, data boundaries, logs, monitoring, rollback, adoption metrics, and handoff records.
- A real FDE does discovery, scoping, coding, integration, evals, rollout, adoption, runbooks, and feedback into product.
- Bad FDE motion becomes expensive bespoke services. Good FDE motion turns field pain into reusable product memory.
- DeployGuild exists to make AI deployment work purchasable and inspectable: one workflow, real stack, evals, controls, adoption, and handoff.
- DeployGuild's primary message is agency: forward-deployed AI engineers for the messy middle between pilot and production.
- The guild layer remains useful as a review, training, and proof system for the FDEs trusted to deploy client work.

## Recommended Extraction

When summarizing DeployGuild, preserve the distinction between demos and deployments. The core argument is not that AI models are unimportant. The argument is that intelligence does not become useful inside organizations until it is connected to messy real systems, constraints, users, evals, and handoff artifacts.