Best of LinkedIn: AI & Agentic Systems CW 43/ 44
The conversation has moved from what agents can do to what they are allowed to do. Governance, identity and evidence now sit ahead of capability, while vendors from Oracle to Microsoft, Google and Snowflake ship the control layers that production deployment requires. The organizations that scale agents will be those that treat them as managed workers inside a redesigned operating model, not as software features.
Date
November 4, 2026
Artificial Intelligence
Thomas Allgeyer

Methodology: Every two weeks we collect most relevant posts on LinkedIn for selected topics and create an overall summary only based on these posts. If you´re interested in the single posts behind, you can find them here: https://linktr.ee/thomasallgeyer. Have a great read!

Listen to our podcast

If you prefer listening, check out our podcast summarizing the most relevant insights from AI & Agentic Systems CW 43/ 44

Governance, Identity and Security

  • Agents are the fastest growing, least governed enterprise identities
  • Combined agent permissions create unplanned data leak paths
  • Passing evaluations is separate from deciding what agents may do
  • 82% of firms run unknown AI agents, widening shadow AI
  • Druva shows scoped permissions and session isolation make agents governable
  • Microsoft finds built in guardrails speed up agent scaling
  • Frontier model governance needs independent validation

‍

Enterprise Architecture and Production Readiness

  • Most enterprise agent failures are architecture failures
  • Framework selection in 2026 is an architecture decision
  • Agent reliability depends on the systems around the model
  • Harness optimization can cut token costs by up to 70%
  • Microsoft Foundry's observability kit catches silent agent failures
  • The A2A protocol standardizes how agents delegate tasks
  • Multi vendor collective intelligence is the next challenge

‍

Platform and Vendor Launches

  • Oracle launches the AI Data Platform Agent Hub and Agent Memory
  • Oracle introduces Fusion Companion and Agentic Home
  • Microsoft moves its Copilot harness into Foundry
  • Google launches a Connected Apps Store and open sources AX
  • NVIDIA, Snowflake and SentinelOne build agent safety into the stack
  • Horizon Context, Web IQ and Zoom Decisions APIs supply agent context
  • Nokia unveils a sovereign agentic AI platform for telco networks

‍

Finance, ERP and Regulated Industries

  • Agentic finance depends on a strong ERP system of record
  • Accounting agents need evidence before gaining posting authority
  • Agents that spend money need independent receipts and controls
  • China's principles make banks the authorization layer for agents
  • A proposed Chase agent card centers on owner defined guardrails
  • Clinical agents need harnesses and clinician review

‍

Workforce and Operating Models

  • Managers become governors of decision systems
  • Agents need an HR like management structure
  • Outdated human processes are the real bottleneck
  • A 2040 finance organization keeps humans on top and agents in the middle
  • Copilot adoption fails without a people focus
  • Agentic recruiting depends on trust, not automation speed

‍

Applied Deployments

  • Agents co wrote 94% of commits in a two month enterprise build
  • Agentic software development becomes a continuous loop under guardrails
  • A multi agent investment system moves from paper to live trading
  • Agents simplify NVIDIA Jetson edge deployments
  • A Slack AI employee turned LinkedIn analytics into a content strategy

‍

Research and Market Data

  • Splunk finds observability essential for trusted agents
  • BCG finds agents framed as employees cut error detection by 18%
  • BCG finds half of CEOs tie job stability to AI success
  • 80% of UK financial firms deploy agents, only 2% allow broad autonomy
  • Shared models make agents hard to insure
  • Research maps agentic inequality across availability, quality and quantity

‍

Subscribe to newsletter

Subscribe to receive the latest blog posts to your inbox every week.

Please confirm your GDPR consent to join our mailing list.
By subscribing you agree to with our Privacy Policy.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Want to see the posts voices behind this summary?

This week’s roundup (CW 43/ 44) brings you the Best of LinkedIn on AI & Agentic Systems

→ 70 handpicked posts that cut through the noise

→ 36 fresh voices worth following

→ 1 deep dive you don’t want to miss

‍