If you have ever tried to add capacity to a software project on short notice, you know the drill: post the role, interview candidates, wait for someone to ramp up on your codebase and your context, and hope the new person clicks with the rest of the team before the deadline arrives anyway. That is the traditional playbook for “team augmentation.” Over the past week, three of the biggest names in enterprise software quietly rewired a different part of that equation. Instead of adding people, they made it dramatically easier to add AI agents directly into the chat channels, meetings, and tools your team already uses. For any business owner who has felt the pinch of scaling a delivery team, that shift is worth understanding now, before it becomes the default way software gets built.

Coding Agents Moved Into the Room Where the Work Already Happens

On August 20, Salesforce launched Slack Code, which lets a team tag an AI coding agent directly inside a Slack conversation. The agent opens a dedicated “code channel,” does the work in the open, and posts back the code changes, a running plan, and a live preview for anyone in the channel to watch and steer. Slack’s chief marketing officer, Ryan Gavin, described the old way of working as “a solo player sport” where an individual’s speed never scaled up to the rest of the organization. The new model is designed to let non-engineers, from marketing to product, watch a feature get built in real time and weigh in before it ships, rather than seeing it only after the fact.

A day later, GitHub shipped a close cousin of the same idea for Microsoft Teams. Mentioning “@GitHub” in a Teams channel, thread, or direct message now starts a Copilot coding agent session that the whole conversation can see and help direct. GitHub’s own example is telling: a team can hand a problem to the agent in the middle of a standup meeting and have it start investigating before the meeting even ends. Anyone with repository access can trigger the agent to make changes, and repository administrators can require an extra layer of human approval before any agent-authored pull request merges.

Why Two Rivals Built the Same Feature in the Same Week

Slack and Microsoft Teams are the two dominant places knowledge workers already spend their day. Neither Salesforce nor GitHub is trying to convince your team to adopt a new tool; they are trying to make sure the AI agent shows up inside the tool your team never left. That is a meaningfully different pitch than the last two years of standalone AI coding assistants, which generally lived in a developer’s terminal or IDE and stayed invisible to everyone else. When the work happens in a shared channel instead of a private tab, a project manager or business owner can watch progress the way they would watch a contractor’s status updates, without learning a new interface.

That visibility matters more than it might sound. One of the quieter frustrations of hiring outside developers is the lag between “the work is happening” and “I can see evidence of it.” Bringing agents into shared channels compresses that gap to nearly zero, which is exactly what an anxious business owner watching a deadline usually wants.

The Plumbing Behind the Scenes Is Catching Up Too

A less visible but arguably more important update landed on August 22, when the maintainers of the Model Context Protocol, the open standard that lets AI agents connect to outside tools and data, published an updated roadmap. A central priority going forward is agent identity: giving an AI agent that is acting on a company’s behalf, without a person actively supervising it in the moment, a verifiable identity and a defined, limited scope of authority, built on established security standards rather than a shared password or an all-access API key.

This is the unglamorous infrastructure work that determines whether visible features like Slack Code or a Copilot session in Teams are something a business can trust with real systems and real data. An agent with a clear, auditable identity and narrowly scoped permissions is a fundamentally different risk proposition than one authenticated with a shared credential that could touch anything. As more vendors race to embed agents into collaboration tools, the standards for who that agent is and what it can touch matter just as much as what it can build.

What This Means If You’re Scaling a Delivery Team

If you are weighing whether to hire, contract, or outsource your next round of development capacity, these shifts are worth turning into direct questions for any software partner you are evaluating:

Where does the AI-assisted work actually happen, and can you see it? A partner whose agents work invisibly in a private tool is harder to audit than one who can show you a live channel, a diff, and a running plan as the work progresses.

What identity and permissions does an agent get, and who approves its output before it ships? Ask specifically whether agent-authored changes require a documented human review step, and whether that step is enforced by the tooling or left to individual habit.

Is “augmenting the team” still assumed to mean adding headcount, or has the plan already accounted for agentic capacity? A delivery plan built entirely around hiring more people may be more expensive and slower than one that has already worked out where agents can safely absorb routine work.

The Takeaway

Team augmentation used to mean one thing: finding more qualified people, fast. That is still true, but it is no longer the only lever available. The last week’s announcements show the industry’s biggest platforms converging on the same idea: put AI agents where the team already talks, give them a verifiable identity, and keep a human in the approval loop. For a business owner, the practical upshot is that “adding capacity” to a project is starting to look less like a hiring decision and more like a configuration decision, provided the delivery partner running it actually understands the difference between visibility and vagueness, and between convenience and control.

If you are trying to figure out how agentic tools fit into your own delivery plans, or whether your current team augmentation approach is keeping pace with what is now possible, the team at Kode Vox can help you think it through. Reach out at info@kodevox.com or visit our contact page to start a conversation.

Sources and further reading:

• Salesforce introduces Slack Code to bring agentic team coding into the open — SiliconANGLE (August 20, 2026)
• Introducing Slack Code: Agentic Coding for Teams — Salesforce
• Shared agentic work with GitHub Copilot in Microsoft Teams — GitHub Changelog (August 21, 2026)
• The New MCP Roadmap — Model Context Protocol Blog (August 22, 2026)

— The Kode Vox Team