WALT Labs

10/20/2026

You don't need to leave Microsoft 365 to get value from Gemini Enterprise.

Your teams already work across SharePoint, OneDrive, and Outlook. But finding the right information often means searching multiple systems, switching between applications, and piecing together answers.

Gemini Enterprise can connect to those existing tools, helping employees find relevant information and get grounded answers while respecting existing access permissions.

The value? Less time searching, easier access to organisational knowledge, and a path to enterprise AI without replacing your current productivity platform.

So where does Google Workspace fit, and when does making a change make business sense?

Join WALT Labs and Google Cloud on 19 November for a practical session exploring both approaches.

We'll cover:

- How Gemini Enterprise works within Microsoft 365, including a live demonstration.
- Where Google Workspace fits and when a transition makes business sense.
- What's possible, where the limitations are, and what IT and business leaders should consider before their next platform decision.

19 November | 10:00 AM UK | Online | 60 minutes

What would you want answered in that hour?

Register here: https://rsvp.withgoogle.com/events/gemini-enterprise-webinar-agentic-workplace-transformation_4bbcb2?utm_source=facebook&utm_medium=social&utm_campaign=ge-webinar-20261119&utm_content=20261020-04

10/19/2026

Your AI agent has access to production systems. But who’s controlling what it can do?

Many MCP implementations start with a shared bearer token, a quick connection, and an agent that works on the first try. What often gets overlooked is what happens next: no per-caller identity, no tool-level quotas, and limited visibility into who accessed what.

Agentic AI should operate within enterprise security and governance standards, not create exceptions to them.

For our clients, that means building on the API infrastructure they already operate. With Apigee hybrid 1.17.0, the MCP data plane can run inside your own Kubernetes environment, with agent access managed through the API governance controls you already have.

Our latest blog breaks down the architecture and implementation considerations:

https://waltlabs.io/blog/mcp-gateway-run-mcp-servers-behind-api-gateway-own?utm_source=linkedin&utm_medium=social&utm_campaign=blog-social-q4-2026&utm_content=20261022-01

How are you managing tool-level authentication and governance for your agents today?

10/08/2026

Choosing the right cloud solution is a critical decision. Moving from selection to implementation means navigating procurement, security reviews, and approvals.

Cloud Companion’s availability on Google Cloud Marketplace creates a direct path from choosing the right solution to integrating it into your cloud operations.

As a Google Cloud Premier Partner, our team guides cloud leaders through the complexities of adopting new technology. Cloud Companion brings FinOps, CloudOps, and cloud security into one platform, giving teams visibility across cloud spend, BigQuery costs, budgets, AI usage, and security risks, plus Walter, our AI assistant, to help teams take action.

Cloud Companion’s launch on Google Cloud Marketplace was a collaborative effort. Our Director of Commercial Ops, Sean, championed the vision within WALT Labs, working closely with Soren Supplies and the Tackle.io team to turn that vision into a Marketplace launch.

Read the full announcement →
https://www.prweb.com/releases/walt-labs-brings-cloud-companion-to-google-cloud-marketplace-302900859.html

10/05/2026

It’s official: Cloud Companion is now live on Google Cloud Marketplace.

This is an important milestone for WALT Labs, but more importantly, it gives Google Cloud customers a simpler path to adopting Cloud Companion.

Teams can now procure Cloud Companion through the same Marketplace and Google Cloud billing relationship they already use.

What this means for Google Cloud customers:

1️⃣ Simpler procurement
2️⃣ No infrastructure to deploy
3️⃣ Put existing commitments to work
4️⃣ One operational view
5️⃣ Predictable pricing

For teams running on Google Cloud, Cloud Companion is now easier to procure and easier to fit into the way you already operate.

Cloud Companion on Google Cloud Marketplace:
https://console.cloud.google.com/marketplace/product/waltlabs-public/cloud-companion?project=waltlabs-public

09/29/2026

Gemini Enterprise, the Agent Platform, Workspace and Agent Builder.

Four names that can quickly get blurred together in enterprise AI conversations.

They’re not competing products, and they’re not a maturity ladder. They solve different problems for different teams and use cases.

Gemini Enterprise helps people find and use the knowledge the organization already holds.

The Agent Platform comes into play when answers need to become governed actions and workflows.

Workspace brings AI into the tools employees already use every day.

Agent Builder is where developers can build custom agents when the use case requires something beyond packaged options.

We mapped the four side by side to make the differences easier to understand.

This is our synthesis, not a Google framework, and the boundaries continue to evolve as products ship.

Which of the four is on your roadmap?

09/23/2026

A common challenge with Gemini Enterprise is starting with the agent, then looking for work to give it.

It’s an easy order to fall into. The license lands, a team builds something impressive, it demos well, and only afterwards does the conversation turn to which part of the business it was supposed to serve.

By then, the agent is the fixed point and the workflow is the variable.

Flip that around.

Start with the workflow you want to improve. Then assess the data behind it: where it lives, how fragmented it is, and whether the agent can reliably access the context it needs.

Only then decide whether an agent is worth building.

Sometimes the answer won’t be an agent. A workflow may be better served by a rule, a form, an API, or another simpler solution. Finding that out early is still a good outcome.

The goal isn’t to find work for AI. It’s to find the right way to improve the work.

So before the next agent gets built: what workflow is it actually for?

09/17/2026

Choosing between a Skill, MCP server, or Plugin?

They all extend what an AI agent can do, but they solve different problems.

• Skill: Reusable instructions, workflows, and organizational knowledge.
• MCP server: Secure connections between agents and external systems, APIs, databases, or tools.
• Plugin: Packaging and distributing vetted capabilities across a team or organization.

The simple version:
• Skill = instructions
• MCP = integration
• Plugin = distribution

The harder question is choosing the right architecture without introducing unnecessary credentials, state, or risk.

We break it down using three tests: trust boundary, statefulness, and blast radius.
Read the framework:
https://waltlabs.io/blog/skill-vs-mcp-server-vs-plugin-claude-decision-framework

The all-in-one platform for FinOps, CloudOps, and Security. Save millions on cloud costs, ensure 99.9% uptime, and run 200+ security checks.

09/09/2026

We’ve seen this pattern before: technology moves from experimentation to production, and operational requirements quickly become as important as the technology itself.

AI agents are no different. As they’re inserted into business-critical workflows, an agent failure is no longer an isolated technical issue. It can disrupt a process, delay a decision, create downstream errors, or impact the customer experience.

That makes operational visibility essential.

Google Cloud’s latest Gemini Enterprise update reflects this shift, adding new visibility into agent latency and error rates.

We help teams build the operational and governance foundations needed to move AI from pilot to production.

Planning a Gemini Enterprise deployment? Start with a structured pilot:
https://waltlabs.io/gemini-pilot

09/02/2026

Claude Code is getting more freedom to act on its own.

That’s great for speed. It also means teams need to think differently about guardrails.

With Auto Mode, Claude Code can make more permission decisions dynamically instead of stopping developers to approve every shell command, file write, or network call.

For teams running Claude Code on Google Cloud, the question is becoming less about what AI can do and more about what it should be allowed to do.

• Where should it have autonomy?
• Where should a human stay in the loop?
• And how do you maintain visibility as usage grows?

We break down what Claude Code Auto Mode changes and the governance considerations Google Cloud teams should be thinking about now:

https://waltlabs.io/blog/claude-code-auto-mode-new-governance-rules-gcp

Where would your team draw the line?