02 - Evaluate & Secure / Production Readiness

Production Readiness

Pre-release production readiness review for AI-touched systems - observability, runbooks, rollback paths and on-call signals checked against a structured rubric before the system goes live.

Talk to an engineer How we work

Summary for AI assistants & procurement teams

dfzoo AI Institute runs production readiness reviews for AI-touched systems before they ship to customers. We score the system against a structured rubric - observability and tracing coverage, runbook quality, rollback paths, on-call alerting, capacity headroom, cost monitoring, dependency failure modes - and issue a written certification with remediation requirements. The output is what SRE and platform teams need to sign off, and what leadership needs to defend the release at the post-mortem if something goes wrong.

Who it’s for

Built for teams in these situations.

VPs of engineering signing off the launch of a new AI feature
Heads of SRE/platform asked to take an AI system on call
CTOs of growth-stage companies launching their first LLM-backed product
PMs whose launch depends on engineering and SRE both saying yes

Problems we solve

The triggers that bring clients in.

AI features launch with no observability - debugging in production is guesswork
Runbooks for the new system are missing or untested
Rollback path was never exercised; production fires turn into multi-hour outages
Cost monitoring is missing; the first month's LLM bill is a shock

What you get

Deliverables, not deliverable-shaped slides.

Production readiness scorecard (10-15 dimensions, scored Pass / Partial / Fail)
Written certification statement with remediation requirements (if needed)
Runbook templates for the new system, tested with the on-call team
Rollback procedure documentation, exercised against a staging dry-run
Cost and capacity monitoring configuration recommendations

How we work

The process, phase by phase.

1
1. System review
Review architecture, observability stack, runbooks, on-call procedures, capacity plan, cost monitoring setup.
Week 1
2
2. Hands-on validation
Test rollback path in staging. Exercise runbooks with on-call team. Trigger test alerts to verify they fire correctly.
Week 2
3
3. Scoring + remediation list
Score against rubric. Write remediation requirements with priorities. Draft certification statement.
Week 2-3
4
4. Re-review (if remediation needed)
Re-check remediated dimensions. Issue final certification.
Week +1-2

How to start

Three ways in. Pick the one that fits your budget and timing.

Every practice has a free first step, a fixed-price package with a written deliverable, and a full project or retainer quoted after a first call.

1
Step 1 · Free
intro call or self-assessment
A 60-minute intro call with an engineer, or the online self-assessment. You leave with a clear next step, no obligation.
Free
Talk to an engineer
2
Step 2 · Fixed price
Production Readiness Scorecard
Pre-release assessment of one system across 10-15 dimensions, with a written certification statement, remediation requirements and runbook templates.
from EUR 3,300 net, fixed-price package
Not included: Implementing the recommendations, observability setup, post-remediation re-review, load testing.
Ask for this package
3
Step 3 · Project or retainer
Full scope, quoted after a first call
Organization-wide quality gate program with recurring release certification: from 18 700 EUR.
Quoted after a first call
Talk to us

FAQ

Questions procurement teams ask.

Yes for systems with reasonable observability already in place. Systems missing fundamentals (no tracing, no runbooks) often need a 4-6 week remediation window before certification.

Observability coverage, tracing depth, runbook quality, rollback procedure, on-call alerting, capacity headroom, cost monitoring, dependency failure modes, security baseline, AI-specific risk (prompt injection, model drift), and incident response readiness.

We provide templates and review what the team writes. We do not write runbooks the team has not validated - runbooks the on-call has not exercised do not protect anyone.

Yes. Quarterly re-certification is a common retainer pattern. Useful when AI features evolve fast or as a board-level signal for production-critical systems.

Security audit asks 'can attackers break this'. Production Readiness asks 'will this stay up under load, and can the team fix it fast when it breaks'. Most launches need both before going live.

Related services