AI Developer Certificate

Directory / Google Cloud

Professional Agentic Architect

A new professional exam on designing and running agentic systems on Google Cloud. It pairs a proctored exam with hands-on labs, a first for Google Cloud's professional tier.

What it signals

Practical assessment: Performance-based labs or a practical build (4 of 4) Practical assessment
Performance-based labs or a practical build. Requires hands-on labs in a live environment in addition to the proctored exam (lab proctoring and details unpublished).
Identity assurance: Proctored with identity verification (4 of 4) Identity assurance
Proctored with identity verification. Part 1: online-proctored or onsite-proctored via Pearson. Part 2: hands-on labs taken in Google Skills; whether labs are proctored is not stated.
Depth: Professional (3 of 4) Depth
Professional. Professional level.
Experience expected: 2 to 3 years (3 of 4) Experience expected
2 to 3 years. Google recommends 3+ years cloud experience including 1+ year building agentic solutions.
Portability: Mixed (2 of 4) Portability
Mixed. Agent patterns, RAG, MCP and A2A are open concepts, but most objectives name Google-specific tooling.
Verifiability: Public badge or registry (4 of 4) Verifiability
Public badge or registry. Google Cloud certifications are issued as public Credly badges; assumed to apply at GA (not confirmed for beta).

For hiring managers

  • Brand new: as of 2026-09-27 nobody can hold this credential yet - beta results come 4-6 weeks after the windows close. A CV claiming it before roughly Nov 2026 should be questioned.
  • First Google Cloud professional certification to combine a proctored exam with hands-on labs.
  • Validity is only 1 year (vs 2 for other professional certs, 3 for foundational).
  • Covers coding agents explicitly, including third-party tools running on Google Cloud, MCP and A2A protocols.

Verifying it

Expected to follow the standard Google Cloud Credly badge process, but not confirmed for this beta credential; no credentials will be issued until beta results are released (4-6 weeks after windows close).

What it covers

Published exam outline

  • Building agents using low-code tools 13%
  • Using coding agents for application development 17%
  • Developing custom agents 33%
  • Evaluating and deploying agentic workflows 22%
  • Securing and governing agentic workflows 15%

By topic

GenAI apps: Substantial (20 to 35%) (3 of 4) GenAI apps
Substantial (20 to 35%). Model selection, RAG pipelines/vector retrieval, prompt templates and evaluation against golden datasets span sections 1, 3 and 4, roughly 20-35%.
Agents: Core focus (over 35%) (4 of 4) Agents
Core focus (over 35%). The entire blueprint is about building, orchestrating, deploying and governing agents.
Classical ML: Not covered (0 of 4) Classical ML
Not covered. No model training or tuning objectives.
Operations: Substantial (20 to 35%) (3 of 4) Operations
Substantial (20 to 35%). Section 4 (22%) covers continuous evaluation, runtime selection, deployment, troubleshooting and monitoring of agents.
Data: Touched on (under 10%) (1 of 4) Data
Touched on (under 10%). Only connecting enterprise data and ingesting unstructured multimodal data (objective 1.2) touch data preparation.
Responsible AI: Some (10 to 20%) (2 of 4) Responsible AI
Some (10 to 20%). Section 5 (15%) covers agent security, governance, guardrails and human-in-the-loop.

Exam details

Who the provider says it is for. 'A technical practitioner who designs and manages autonomous, AI-driven agentic workflows in Google Cloud' - an experienced developer or architect with deep experience using LLMs, applying agent design patterns, writing code and integrating data sources.

Format
multiple choice (proctored, Pearson); hands-on labs (in Google Skills), unlocked after passing the multiple-choice part
Delivery
Part 1: online-proctored or onsite-proctored via Pearson. Part 2: hands-on labs taken in Google Skills; whether labs are proctored is not stated.
Code or hands-on work
Yes. Two-part assessment: proctored multiple-choice exam plus hands-on labs in Google Skills. Role description requires 'writing code'. Number, length and scoring of the labs are not published.
Prerequisites
None
Recommended experience
3+ years hands-on experience building, testing, deploying and managing cloud solutions, including 1+ years building agentic solutions using Google Cloud
Price notes
Beta fee is USD 120 (40% off the USD 200 retail price, per official page). Vouchers accepted.

Preparing

Not confirmed

These details were not available from an official source at the time of research.

  • exact beta registration opening date (third-party gives both 3 and 8 Sept 2026)
  • GA date and GA exam length/question count
  • number, duration, proctoring and pass criteria of hands-on labs
  • passingScore
  • renewal method
  • whether badge is issued via Credly
  • learning path page returned HTTP 403 to curl but loaded via WebFetch
Sources (4)