Back to FAQ Hub
FAQ Category

Delivery & Handover

What you receive at production cutover, how source code ownership works, and what the 6-month QA window covers.

Overview

Handover is where vendor lock-in is decided. DevStudio delivers full source code, infrastructure-as-code, the eval set, prompts, deployment docs, and a runbook to your own accounts from day one — there are no vendor-locked layers. This page covers exactly what you own at cutover, what the 6-month QA window includes, how model upgrades are handled, and what happens during a critical incident.

Key things to know about delivery

You own everything at cutover

Full source code in your GitHub org from commit one, infrastructure-as-code in your cloud accounts, the eval set, prompts, orchestration logic, deployment docs, and a runbook. You can replace us at any milestone and continue.

The 6-month QA window is concrete

It covers production fixes, quarterly Token Audits, and business-hours on-call for production-impacting issues. Net-new features and model upgrades are separate, clearly-scoped change requests.

Model upgrades are gated by the eval set

On every upgrade we re-run the full eval set on the new model and only cut traffic over once accuracy recovers, so you are not surprised by silent regressions.

Frequently Asked Questions

What do I own at handover?
Full source code (your GitHub organization from commit one), infrastructure-as-code (your cloud accounts), eval set, prompts and orchestration logic, deployment docs, runbook with top 12 failure classes, and on-call escalation path. There are no vendor-locked layers; you can replace us at any milestone and continue from where we left off. See /blog/source-code-ownership-in-outsourcing-projects/ for the contract language.
What does the 6-month QA window cover?
Production fixes (any reported defect rooted in the engagement deliverable), quarterly Token Audits (routing, caching, model selection re-evaluated against the eval set), and on-call response during business hours for production-impacting issues. Out of scope: net-new features, scope additions, model upgrades — those are separate change-request engagements.
What happens after the 6-month QA window?
Two paths. Operate-with-you retainer (monthly cost + scope, typically $5k-$25k/month depending on workload) where we keep tuning. Or full handover where your in-house team owns ops and we are available for paid escalations on hourly basis. The handover plan is agreed at engagement signature, not at month 6.
How do model upgrades get handled?
On every model upgrade, we re-run the full eval set on the new model. If accuracy drops, we tune prompts, retrieval, or thresholds until accuracy recovers, and only then cut traffic over. Buyers who skip this step are the ones surprised by silent regressions when a provider ships a new version. Included in the operate-with-you retainer.
What if a critical incident happens at 2 AM?
Production agents ship with an on-call runbook covering the top 12 failure classes, alerting tuned to the SLO, and a one-click rollback to the previous safe state. For buyers on the operate-with-you tier, we provide a defined incident response time tied to severity. For buyers who own ops fully, the runbook is written so your existing on-call team can triage without us. See /blog/how-to-accept-an-ai-outsourcing-project/ for the full handover checklist.