Build with AI
on Google Cloud
Session #4 AI Agents
3/5/2025
Seattle | Surrey | Vancouver | Burnaby
GDG Seattle
2
Margaret Maynard-Reid
Yenchi Lin
Clive Boulton
Vishal Pallerla
I/O Extended 2019
2024 Build with AI
DevFest Seattle 2018
DevFest Seattle 2022
WTM Lightning Talks 2018
Cloud Study Jam 2018
DevFest Seattle 2017
DevFest Seattle 2016
DevFest Seattle 2015
DevFest Seattle 2024
Follow GDG Seattle on LinkedIn
GDG Surrey
3
Follow GDG Surrey on LinkedIn
GDG Vancouver
Follow GDG Seattle on LinkedIn
4
Follow GDG Vancouver on LinkedIn
Join our GDG Vancouver Community
Volunteer Interest Form
GDG Burnaby
GDG Burnaby Bevy | LinkedIn
5
Build with AI
on Google Cloud
Agenda
● Study series overview
● Talk 1: Jason Davenport, Google
● Talk 2: Henry Ruiz, ML GDE
● Q & A
Seattle | Surrey | Vancouver | Burnaby
Build with AI
on Google Cloud
Study series
overview
7
8
Link to Story on Medium
Study series overview
Follow 5 generative AI paths on Google Cloud Skills Boost:
1. 1/22/25 - Beginner: Intro to GenAI (link)
2. 2/5/25 - Generate Smarter GenAI Outputs (link)
3. 2/19/25 - Build & Modernize Apps with GenAI (link)
4. 3/5/25 - Integrate GenAI into Your DataFlow (link)
5. 3/19/25 - Deploy & Manage GenAI Models (link)
Session topics are not limited to the above.
Each session: 2 short talks (by Googlers or experts) + Q&A section.
9
What is a CSB learning path?
A learning path has multiple courses
Each course has videos, recommended reading, quiz & hands-on labs.
You will have at least two weeks to work through the materials
It’s OK if you don’t finish and feel free to study ahead
10
Access to Cloud Skills Boost
● Sign up here: https://www.cloudskillsboost.google/
● By RSVP, you get free access for a few months
● The videos are accessible by default while labs each require a credit
● You can work on each GenAI paths before or after each session
Note: Make sure to sign up on Google Cloud Skills Boost with the same email that you used
for event RSVP.
11
Build with AI
on Google Cloud
Google Cloud
meets your agents
Jason Davenport
Tech Lead,
Google Cloud DevRel
12
Proprietary & Confidential
Agents present the biggest opportunity to drive business
impact with AI
Proprietary + Confidential
AI Agents plan, reason, and execute tasks for users
Generative AI Models
(An agent can use
multiple models)
APIs Functions
Databases Agents
Tools
Profile, goals &
instructions
short-term long-term
Memory
Model based
Reasoning/Planning
(Question Decomp &
Reflection)
End User
query
Agent Runtime
Orchestration
(e.g. Agent Brain)
response
Four Key Components
● Model: Used to reason over
goals, determine the plan and
generate a response
● Tools: Fetch data, perform
actions or transactions by
calling other APIs or services
● Orchestration: Maintain
memory and state (including
the approach used to plan),
tools, data provided/fetched,
etc
● Runtime: Execute the system
when invoked
Enterprise
Governance
Gemini API
Agent “inner-loop” &
built-in tools
(code-interpreter, search, etc)
Gemini Models
Optimized for agentic reasoning
and planning
Agent SDK
Agent Engine (Fully-managed runtime)
Business Users
Build agents without code
Tools and
Integrations
Agent “outerloop” orchestration
Trace
Logging
Example Store
Monitoring
Agentic
Autorater
Session
Management
Long-term
memory
Multimodal Intelligence Engine
Agentspace
Enterprise Connectors Knowledge Graph
Conversational Agents
part of Customer Engagement Suite
No-code agents, fully
managed runtime
Mix deterministic
flows and LLMs
Integrated analytics
and CICD
Platform integrations
Developer Platform
Develop agents with desired level of
abstraction
Off-the-shelf (SaaS)
Use Agents
Demo!
https://github.com/GoogleCloudPlatform/agent-starter-pack
Build with AI
on Google Cloud
17
Building a Job Matching
Tool with Multi-Verbal
Agent Collaboration
Henry Ruiz
Research Scientist,
Texas A&M University
LLMs as "reasoning engines"
LLM can model and understand natural language
https://www.promptingguide.ai/research/llm-reasoning
https://arxiv.org/pdf/2404.01230
Agents (An unified solution
for reasoning and Act)
Agents extend beyond text generation
(standalone LLM interaction),
incorporating advanced cognitive
functions such as reasoning,
planning, and decision-making.
By harnessing the advanced
capabilities of large language
models, agents can autonomously
execute complex tasks and solve
multifaceted problems.
Multi-agent and multi-modal applications
represent the state of the art
● Involve a group of agents with
diverse capabilities, including
language models and various tools.
● Agents work collaboratively to
address and solve complex tasks.
● Each agent may specialize in
different aspects of the problem,
contributing to the system's overall
effectiveness and efficiency.
23
Code demo!
Build with AI
on Google Cloud
Q&A
24
Build with AI
on Google Cloud
Cloud Skills Boost
walkthrough
25
26
Sign In -Google Cloud Skills Boost
27
Explore Paths
More questions?
Post them on GDG Surrey
Discord server #gen_ai_gcp
28
Scan Me
Have fun studying!
Action items:
● Join discord - post your questions there
● Get access to Cloud Skills Boost credits
● Complete 4th GenAI path on CSB
● Get started on 5th GenAI path on CSB
Next session:
● Mar 19, 2025 - Session #5 MLOps (RSVP)
29
Upcoming events
Build with AI (Seattle) on 3/18/2025 - organized by Google Cloud
GDG Seattle cross-post Bevy page:
https://gdg.community.dev/events/details/google-gdg-seattle-presents-cross-
post-build-with-ai-hosted-by-google-cloud/
RSVP on the official event website:
https://rsvp.withgoogle.com/events/build-with-ai-seattle
30

Build with AI on Google Cloud Session #4

  • 1.
    Build with AI onGoogle Cloud Session #4 AI Agents 3/5/2025 Seattle | Surrey | Vancouver | Burnaby
  • 2.
    GDG Seattle 2 Margaret Maynard-Reid YenchiLin Clive Boulton Vishal Pallerla I/O Extended 2019 2024 Build with AI DevFest Seattle 2018 DevFest Seattle 2022 WTM Lightning Talks 2018 Cloud Study Jam 2018 DevFest Seattle 2017 DevFest Seattle 2016 DevFest Seattle 2015 DevFest Seattle 2024 Follow GDG Seattle on LinkedIn
  • 3.
    GDG Surrey 3 Follow GDGSurrey on LinkedIn
  • 4.
    GDG Vancouver Follow GDGSeattle on LinkedIn 4 Follow GDG Vancouver on LinkedIn Join our GDG Vancouver Community Volunteer Interest Form
  • 5.
    GDG Burnaby GDG BurnabyBevy | LinkedIn 5
  • 6.
    Build with AI onGoogle Cloud Agenda ● Study series overview ● Talk 1: Jason Davenport, Google ● Talk 2: Henry Ruiz, ML GDE ● Q & A Seattle | Surrey | Vancouver | Burnaby
  • 7.
    Build with AI onGoogle Cloud Study series overview 7
  • 8.
    8 Link to Storyon Medium
  • 9.
    Study series overview Follow5 generative AI paths on Google Cloud Skills Boost: 1. 1/22/25 - Beginner: Intro to GenAI (link) 2. 2/5/25 - Generate Smarter GenAI Outputs (link) 3. 2/19/25 - Build & Modernize Apps with GenAI (link) 4. 3/5/25 - Integrate GenAI into Your DataFlow (link) 5. 3/19/25 - Deploy & Manage GenAI Models (link) Session topics are not limited to the above. Each session: 2 short talks (by Googlers or experts) + Q&A section. 9
  • 10.
    What is aCSB learning path? A learning path has multiple courses Each course has videos, recommended reading, quiz & hands-on labs. You will have at least two weeks to work through the materials It’s OK if you don’t finish and feel free to study ahead 10
  • 11.
    Access to CloudSkills Boost ● Sign up here: https://www.cloudskillsboost.google/ ● By RSVP, you get free access for a few months ● The videos are accessible by default while labs each require a credit ● You can work on each GenAI paths before or after each session Note: Make sure to sign up on Google Cloud Skills Boost with the same email that you used for event RSVP. 11
  • 12.
    Build with AI onGoogle Cloud Google Cloud meets your agents Jason Davenport Tech Lead, Google Cloud DevRel 12
  • 13.
    Proprietary & Confidential Agentspresent the biggest opportunity to drive business impact with AI
  • 14.
    Proprietary + Confidential AIAgents plan, reason, and execute tasks for users Generative AI Models (An agent can use multiple models) APIs Functions Databases Agents Tools Profile, goals & instructions short-term long-term Memory Model based Reasoning/Planning (Question Decomp & Reflection) End User query Agent Runtime Orchestration (e.g. Agent Brain) response Four Key Components ● Model: Used to reason over goals, determine the plan and generate a response ● Tools: Fetch data, perform actions or transactions by calling other APIs or services ● Orchestration: Maintain memory and state (including the approach used to plan), tools, data provided/fetched, etc ● Runtime: Execute the system when invoked
  • 15.
    Enterprise Governance Gemini API Agent “inner-loop”& built-in tools (code-interpreter, search, etc) Gemini Models Optimized for agentic reasoning and planning Agent SDK Agent Engine (Fully-managed runtime) Business Users Build agents without code Tools and Integrations Agent “outerloop” orchestration Trace Logging Example Store Monitoring Agentic Autorater Session Management Long-term memory Multimodal Intelligence Engine Agentspace Enterprise Connectors Knowledge Graph Conversational Agents part of Customer Engagement Suite No-code agents, fully managed runtime Mix deterministic flows and LLMs Integrated analytics and CICD Platform integrations Developer Platform Develop agents with desired level of abstraction Off-the-shelf (SaaS) Use Agents
  • 16.
  • 17.
    Build with AI onGoogle Cloud 17 Building a Job Matching Tool with Multi-Verbal Agent Collaboration Henry Ruiz Research Scientist, Texas A&M University
  • 18.
    LLMs as "reasoningengines" LLM can model and understand natural language https://www.promptingguide.ai/research/llm-reasoning https://arxiv.org/pdf/2404.01230
  • 19.
    Agents (An unifiedsolution for reasoning and Act) Agents extend beyond text generation (standalone LLM interaction), incorporating advanced cognitive functions such as reasoning, planning, and decision-making. By harnessing the advanced capabilities of large language models, agents can autonomously execute complex tasks and solve multifaceted problems.
  • 21.
    Multi-agent and multi-modalapplications represent the state of the art ● Involve a group of agents with diverse capabilities, including language models and various tools. ● Agents work collaboratively to address and solve complex tasks. ● Each agent may specialize in different aspects of the problem, contributing to the system's overall effectiveness and efficiency.
  • 23.
  • 24.
    Build with AI onGoogle Cloud Q&A 24
  • 25.
    Build with AI onGoogle Cloud Cloud Skills Boost walkthrough 25
  • 26.
    26 Sign In -GoogleCloud Skills Boost
  • 27.
  • 28.
    More questions? Post themon GDG Surrey Discord server #gen_ai_gcp 28 Scan Me
  • 29.
    Have fun studying! Actionitems: ● Join discord - post your questions there ● Get access to Cloud Skills Boost credits ● Complete 4th GenAI path on CSB ● Get started on 5th GenAI path on CSB Next session: ● Mar 19, 2025 - Session #5 MLOps (RSVP) 29
  • 30.
    Upcoming events Build withAI (Seattle) on 3/18/2025 - organized by Google Cloud GDG Seattle cross-post Bevy page: https://gdg.community.dev/events/details/google-gdg-seattle-presents-cross- post-build-with-ai-hosted-by-google-cloud/ RSVP on the official event website: https://rsvp.withgoogle.com/events/build-with-ai-seattle 30