�u�}���`�G�[�W�F���g�V�X�e���v�݌v�p�^�[��8�I�@Google������F�u�}�C�N���T�[�r�X�A�[�L�e�N�`����AI�Łv�ŐM��������Ɋ�^

Google�́A�}���`AI�G�[�W�F���g�V�X�e���ɂ�����8�‚̊�{�I�ȃf�U�C���p�^�[���������u���O�ʼn�������B

» 2026�N01��27�� 13��00�� ���J
[��IT]

���̋L���͉������ł��B����o�^�i�����j����ƑS�Ă������������܂��B

�@Google��2025�N12��16���i�č����ԁj�Ɍ����u���O�ŁA�}���`AI�i�l�H�m�\�j�G�[�W�F���g��8�‚̊�{�I�ȃf�U�C���p�^�[������������B

�@Google�́u���m���V�b�N�A�v���P�[�V�������X�P�[���u���ł͂Ȃ��悤�ɁA�P���AI�G�[�W�F���g�́w��p�n�R�x�Ɋׂ�v�Ƃ��A�}���`�G�[�W�F���g�V�X�e���𗘗p���闘�_�́A�u�}�C�N���T�[�r�X�A�[�L�e�N�`����AI�ł��\�z���邱�ƂŁA���U���Ɛ�剻�ɂ��M�����������ł��邱�Ɓv�������Ă���B

�@�u�p�[�T�i��͊�j�A�N���e�B�b�N�i��]�Ɓj�A�f�B�X�p�b�`���[�i���z�ҁj�Ƃ���������̖������X�̃G�[�W�F���g�Ɋ��蓖�Ă邱�ƂŁA�{���I�Ƀ��W���[��������A���؉”\�ŐM�����̍����V�X�e�����\�z�ł���v�iGoogle�j

�@Google�́A8�‚̃}���`�G�[�W�F���g�f�U�C���p�^�[���ɂ‚��āA�I�[�v���\�[�X��AI�G�[�W�F���g�J���^�f�v���C�t���[�����[�N�uAgent Development Kit�v�iADK�j��Python��p�����^���R�[�h�ƕ����Ď��̂悤�ɏЉ�Ă���B

1. �V�[�P���V�����p�C�v���C���p�^�[���i�g�ݗ��ă��C���j

�@�G�[�W�F���g���[�N�t���[�̊�{���̊�{�ł���A�f�[�^�����p�C�v���C���̒�ԃA�[�L�e�N�`�����B

�@�ÓT�I�ȑg�ݗ��ă��C���̂悤�ɁA�G�[�W�F���gA���^�X�N���������A���̃o�g�����G�[�W�F���gB�ɒ��ړn���B�����I�Ō���_�I�ł���A�f�[�^�̏o�ǂ��낪��ɖ��m�Ȃ��߁A���Ƀf�o�b�O�����₷���B

�@�ȉ��̗�́A���̃h�L�������g�̏����t���[�������Ă���B�uParser�v�i��́j�G�[�W�F���g��PDF���e�L�X�g�ɕϊ����A�uExtractor�v�i���o�j�G�[�W�F���g���\�����f�[�^�𒊏o���A�uSummarizer�v�i�v��j�G�[�W�F���g���ŏI�I�ȗv��𐶐�����B

V[PVpCvCG[WFgp^[i񋟁FGooglej �V�[�P���V�����p�C�v���C���G�[�W�F���g�p�^�[���i�񋟁FGoogle�j
# ADK Pseudocode
# Step 1: Parse the PDF
parser = LlmAgent(
    name="ParserAgent",
    instruction="Parse raw PDF and extract text.",
    tools=[PDFParser],
    output_key="raw_text" 
)
# Step 2: Extract structured data
extractor = LlmAgent(
    name="ExtractorAgent",
    instruction="Extract structured data from {raw_text}.",
    tools=[RegexExtractor],
    output_key="structured_data"
)
# Step 3: Summarize
summarizer = LlmAgent(
    name="SummarizerAgent",
    instruction="Generate summary from {structured_data}.",
    tools=[SummaryEngine]
)
# Orchestrate the Assembly Line
pipeline = SequentialAgent(
    name="PDFProcessingPipeline",
    sub_agents=[parser, extractor, summarizer]
)

2. �R�[�f�B�l�[�^�[�^�f�B�X�p�b�`���[�p�^�[���i�R���V�F���W���j

�@�`�F�[���i�A���j�ł͂Ȃ��A�ӎv����҂��K�v�ȏꍇ������B���̃p�^�[���ł́A�����̃C���e���W�F���g�G�[�W�F���g���f�B�X�p�b�`���[�i���z�ҁj�Ƃ��ċ@�\����B

�@���[�U�[�̈Ӑ}�𕪐͂��A�w�����ꂽ�^�X�N�ɍœK�Ȑ��G�[�W�F���g�Ƀ��N�G�X�g��U�蕪����B

�@����́A���̐}�̂悤�ɁA�������Ɋւ�����Ȃ�uBilling�v�i�����j���G�[�W�F���g�ցA�g���u���V���[�e�B���O�Ȃ�uTech Support�v�i�e�N�j�J���T�|�[�g�j���G�[�W�F���g�ցA���[�U�[��U������K�v������悤�ȁA���G�Ȍڋq�T�[�r�X�ɓK���Ă���Ƃ����B

R[fBl[^[^fBXpb`[G[WFgp^[i񋟁FGooglej �R�[�f�B�l�[�^�[�^�f�B�X�p�b�`���[�G�[�W�F���g�p�^�[���i�񋟁FGoogle�j
# ADK Pseudocode
billing_specialist = LlmAgent(
    name="BillingSpecialist", 
    description="Handles billing inquiries and invoices.",
    tools=[BillingSystemDB]
)
tech_support = LlmAgent(
    name="TechSupportSpecialist", 
    description="Troubleshoots technical issues.",
    tools=[DiagnosticTool]
)
# The Coordinator (Dispatcher)
coordinator = LlmAgent(
    name="CoordinatorAgent",
    # The instructions guide the routing logic
    instruction="Analyze user intent. Route billing issues to BillingSpecialist and bugs to TechSupportSpecialist.",
    sub_agents=[billing_specialist, tech_support]
)

3. ���s����^�W��p�^�[���i�^�R���j

�@�݂��Ɉˑ����Ȃ������̃^�X�N������ꍇ�A��������‚��Ž��s����K�v�͂Ȃ��B���̃p�^�[���ł́A�����̃G�[�W�F���g���^�X�N�𓯎��Ɏ��s���A���C�e���V�i�x���j��Z�k������A���l�Ȏ��_�𓾂��肷��B���̌�A�uSynthesizer�v�i�����j�G�[�W�F���g���o�͂��W�񂷂�B

�@����́A�����R�[�h���r���[�Ȃǂɗ��z�I���B���ԂɃ`�F�b�N���s���̂ł͂Ȃ��A�uSecurity Auditor�v�i�Z�L�����e�B�č��j�A�uStyle Enforcer�v�i�X�^�C�������j�A�uPerformance Analyst�v�i�p�t�H�[�}���X���́j�̊e�G�[�W�F���g�𓯎��ɋN�����A�v�����N�G�X�g�����r���[�����邱�Ƃ��ł���B�S�G�[�W�F���g�����r���[������������A�uSynthesizer�v�i�����j�G�[�W�F���g�������̃t�B�[�h�o�b�N���܂Ƃ߂āA1�‚̃��r���[�R�����g���쐬����B

sG[WFgp^[i񋟁FGooglej ���s����G�[�W�F���g�p�^�[���i�񋟁FGoogle�j
# ADK Pseudocode
# Define parallel workers
security_scanner = LlmAgent(
    name="SecurityAuditor", 
    instruction="Check for vulnerabilities like injection attacks.",
    output_key="security_report"
)
style_checker = LlmAgent(
    name="StyleEnforcer", 
    instruction="Check for PEP8 compliance and formatting issues.",
    output_key="style_report"
)
complexity_analyzer = LlmAgent(
    name="PerformanceAnalyst", 
    instruction="Analyze time complexity and resource usage.",
    output_key="performance_report"
)
# Fan-out (The Swarm)
parallel_reviews = ParallelAgent(
    name="CodeReviewSwarm",
    sub_agents=[security_scanner, style_checker, complexity_analyzer]
)
# Gather/Synthesize
pr_summarizer = LlmAgent(
    name="PRSummarizer",
    instruction="Create a consolidated Pull Request review using {security_report}, {style_report}, and {performance_report}."
)
# Wrap in a sequence
workflow = SequentialAgent(sub_agents=[parallel_reviews, pr_summarizer])

4. �K�w�I�����i�}�g�����[�V�J�j

�@�^�X�N���傫�����āA1�‚̃G�[�W�F���g�̃R���e�L�X�g�E�B���h�E�Ɏ��܂�Ȃ����Ƃ�����B��ʃG�[�W�F���g�́A���G�ȖڕW���T�u�^�X�N�ɕ������A�������ϔC�ł���B�U�蕪���p�^�[���Ƃ͈قȂ�A�e�G�[�W�F���g�̓^�X�N�̈ꕔ�������ϔC���A���̌��ʂ�҂��Ă��玩�g�̐��_���p������ꍇ������B

�@���̐}�́uReportWriter�v�i���|�[�g�쐬�ҁj�́A���g�ł̓��T�[�`���s��Ȃ��B�uResearchAssistant�v�i���T�[�`�A�V�X�^���g�j�ɈϔC���A�����ResearchAssistant�́A�uWebSearch�v�iWeb�����j����сuSummarizer�v�i�v��j�c�[�����Ǘ����Ă���B

KwI^XNG[WFgp^[i񋟁FGooglej �K�w�I�^�X�N�����G�[�W�F���g�p�^�[���i�񋟁FGoogle�j
# ADK Pseudocode
# Level 3: Tool Agents
web_searcher = LlmAgent(name="WebSearchAgent", description="Searches web for facts.")
summarizer = LlmAgent(name="SummarizerAgent", description="Condenses text.")
# Level 2: Coordinator Agent
research_assistant = LlmAgent(
    name="ResearchAssistant",
    description="Finds and summarizes info.",
    # Coordinator manages the tool agents
    sub_agents=[web_searcher, summarizer] 
)
# Level 1: Top-Level Agent
report_writer = LlmAgent(
    name="ReportWriter",
    instruction="Write a comprehensive report on AI trends. Use the ResearchAssistant to gather info.",
    # Wrap the sub-agent hierarchy as a tool for the parent
    tools=[AgentTool(research_assistant)] 
)

5. �����҂Ɣ�]�Ɓi�ҏW�f�X�N�j

�@���i���ŐM�����̍����o�͂𐶐�����ɂ́A�����̏ꍇ�A��O�҂̖ڂ��K�v���B���̃p�^�[���ł́A�R���e���c�̍쐬�ƌ��؂𕪗�����B����̃G�[�W�F���g���uGenerator�v�i�����ҁj�Ƃ��ăh���t�g�i�������j���쐬���A����������uCritic�v�i��]�Ɓj�Ƃ��āA�n�[�h�R�[�h���ꂽ���_���`�F�b�N�Ɋ�Â��Ă�������r���[����B

�@���̃A�[�L�e�N�`���̓����́A�����t�����[�v�ɂ���B���r���[�ɍ��i����΁A���[�v�𔲂��A�R���e���c���m�肷��B���i���Ȃ������ꍇ�́A��̓I�ȃt�B�[�h�o�b�N��Generator�ɖ߂���AGenerator�͊�𖞂����h���t�g�𐶐�����B����́A�\���`�F�b�N���K�v�ȃR�[�h������A�R���v���C�A���X���r���[���K�v�ȃR���e���c�쐬�ɔ��ɗL�p���B

҂Ɣ]ƃG[WFgp^[i񋟁FGooglej �����҂Ɣ�]�ƃG�[�W�F���g�p�^�[���i�񋟁FGoogle�j
# ADK Pseudocode
# The Generator 
generator = LlmAgent(
    name="Generator",
    instruction="Generate a SQL query. If you receive {feedback}, fix the errors and generate again.",
    output_key="draft"
)
# The Critic 
critic = LlmAgent(
    name="Critic",
    instruction="Check if {draft} is valid SQL. If correct, output 'PASS'. If not, output error details.",
    output_key="feedback"
)
# The Loop 
loop = LoopAgent(
    name="ValidationLoop",
    sub_agents=[generator, critic],
    condition_key="feedback",
    exit_condition="PASS" 
)

6. �����I���ǁi�����Ɓj

�@�h���t�g��1�����������ŁA�D�ꂽ���ʕ����������邱�Ƃ͂܂ꂾ�B�l�Ԃ̍�Ƃ����e�𐄝ȁi���������j���A�����グ�A�ҏW����悤�ɁA�G�[�W�F���g���A�����𐳊m�ɂ��邽�߂ɐ���̎��s���K�v�ȏꍇ������B���̃p�^�[���ł́A�o�͂�����̕i����ɒB����܂ŁA�����A��]�A���ǂ̃T�C�N�����J��Ԃ��B

�@�����҂Ɣ�]�ƃp�^�[��������i���i�^�s���i�j�ɏœ_�𓖂Ă�̂ɑ΂��A���̃p�^�[���͎��I�ȉ��P�ɏœ_�𓖂Ă�B

�@�uGenerator�v�i�����ҁj�G�[�W�F���g���e���h���t�g���쐬���A�uCritique�v�i��]�Ɓj�G�[�W�F���g���œK���̂��߂̎w�E���s���A�uRefinement�v�i���ǁj�G�[�W�F���g�������̎w�E�Ɋ�Â��ďo�͂𖁂��グ��B

IǃG[WFgp^[i񋟁FGooglej �����I���ǃG�[�W�F���g�p�^�[���i�񋟁FGoogle�j
# ADK Pseudocode
# Generator
generator = LlmAgent(
    name="Generator",
    instruction="Generate an initial rough draft.",
    output_key="current_draft"
)
# Critique Agent 
critic = LlmAgent(
    name="Critic",
    instruction="Review {current_draft}. List ways to optimize it for performance.",
    output_key="critique_notes"
)
# Refiner Agent 
refiner = LlmAgent(
    name="Refiner",
    instruction="Read {current_draft} and {critique_notes}. Rewrite the draft to be more efficient.",
    output_key="current_draft" # Overwrites the draft with better version
)
# The Loop (Critique -> Refine)
loop = LoopAgent(
    name="RefinementLoop",
    max_iterations=3, 
    sub_agents=[critic, refiner]
)
# Complete Workflow
workflow = SequentialAgent(sub_agents=[generator, loop])

7. �q���[�}���C���U���[�v�i�l�Ԃ̃Z�[�t�e�B�[�l�b�g�j

�@AI�G�[�W�F���g�͋��͂����A�d�v�Ȉӎv����ɂ͐l�Ԃ�������Ȃ���΂Ȃ�Ȃ��B���̃p�^�[���ł́A�G�[�W�F���g����b�I�ȍ�Ƃ��s�����A�d��ȍs���i���ɁA�s�‹t�I�ȍs����A�d��Ȍ��ʂ𔺂��s���j�ɂ‚��ẮA�l�Ԃ����F����K�v������B���S���Ɛ����ӔC���m�ۂ��邽�߂��B�����̍s���ɂ́A���Z����̎��s�A�{�Ԋ‹��ւ̃R�[�h�f�v���C�A�@���f�[�^�Ɋ�Â��s���Ȃǂ��܂܂��B

�@���̐}�́A�uTransaction�v�i�g�����U�N�V�����j�G�[�W�F���g������Ɩ�����������l�q�������B�d��ȃ`�F�b�N���K�v�ɂȂ�ƁA�uApprovalTool�v�i���F�c�[���j�G�[�W�F���g���Ăяo���A���s���ꎞ��~���āA�l�Ԃ̃��r���[�S���҂�����𔻒f����̂�҂B

q[}CU[vG[WFgp^[i񋟁FGooglej �q���[�}���C���U���[�v�G�[�W�F���g�p�^�[���i�񋟁FGoogle�j
# ADK Pseudocode
transaction_agent = LlmAgent(
    name="TransactionAgent",
    instruction="Handle routine processing. If high stakes, call ApprovalTool.",
    tools=[ApprovalTool] 
)
approval_agent = LlmAgent(
    name="ApprovalToolAgent",
    instruction="Pause execution and request human input."
)
workflow = SequentialAgent(sub_agents=[transaction_agent, approval_agent])

8. �����p�^�[���i���܂��܂ȃp�^�[���̑g�ݍ��킹�j

�@�{�ԃO���[�h�̃A�v���P�[�V�������\�z����ɂ́A�����̃p�^�[����g�ݍ��킹��K�v�����邾�낤�B

�@�Ⴆ�΁A���̌��S�i����낤�j�Ȍڋq�T�|�[�g�V�X�e���ł́A���N�G�X�g�̐U�蕪���ɁuCoordinator�v�i�R�[�f�B�l�[�^�[�j���g�p����B

�@���[�U�[���Z�p�I�Ȗ�������Ă���ꍇ�A�uDocs Search�v�i���������j�ƁuUser History�v�i���[�U�[�����j�̊e�G�[�W�F���g�������ɋN������A�������s�ŏ������s���B�ŏI�I�ȉ񓚂́A���[�U�[�ɑ��M�����O�ɁA�g�[���̈�ѐ����m�ۂ��邽�߂ɁA�uGenerator�v�i�����ҁj�ƁuCritic�v�i��]�Ɓj�̃��[�v��ʉ߂���”\��������B

G[WFgp^[i񋟁FGooglej �����G�[�W�F���g�p�^�[���i�񋟁FGoogle�j

Copyright © ITmedia, Inc. All Rights Reserved.

'; }; BodyAdGAMWithCCE.prototype = Object.create(BodyAdContent.prototype); BodyAdGAMWithCCE.EVENT_INIT = 'init'; BodyAdGAMWithCCE.EVENT_READY = 'ready'; /** * AdContent �̏������Ɨ��p���� */ BodyAdGAMWithCCE.prototype.init = function () { Logger.writeln('BodyAdGAMWithCCE.prototype.init'); this.dispatchEvent(new CustomEvent(BodyAdGAMWithCCE.EVENT_INIT)); }; /** * �L���g�� GAM ���N�G�X�g */ BodyAdGAMWithCCE.prototype.activate = function () { refreshGam('InArtSpecialLink'); this.dispatchEvent(new CustomEvent(BodyAdGAMWithCCE.EVENT_READY)); }; // global reference window.itm = itm; //entry point const build = function( e ) { BodyAdEventBase.polyfill(); const bodyAdManager = BodyAdManager.getInstance(); bodyAdManager.addEventListener(BodyAdManager.EVENTS.READY, function (ev) { bodyAdManager.loadAdvertise(); }); bodyAdManager.init(); } build(); })();

�A�C�e�B���f�B�A����̂��m�点

�X�|���T�[����̂��m�点PR

���ڂ̃e�[�}

Microsoft  WindowsőO2026
lɗȂA{ŌZLeBu_iCYv
4AI by IT - AIAAA
AI for GWjAO
[R[h^m[R[h Zg by IT - ITGWjArWlX̒SŊ􂷂gD
Cloud Native Central by IT - XP[uȔ\͂gD
�V�X�e���J���m�E�n�E �y�����i�r�zPR
���Ȃ��ɂ������߂̋L��PR

RSS�ɂ‚���

�A�C�e�B���f�B�AID�ɂ‚���

���[���}�K�W���o�^

��IT�̃��[���}�K�W���́A �������A���ׂĖ����ł��B���Ѓ��[���}�K�W�������w�ǂ��������B