AI work is moving into the places where teams already talk. In August 2026, Slack, DeepSeek, and SuperApp introduced products that put coding agents, software tools, and multiple AI models closer to the same shared workspace.
The common idea is simple: conversations should not stop when the work starts. Instead, the discussion, instructions, code, files, tool calls, feedback, and approvals should stay connected. Each company is taking that idea in a different direction, from project channels to autonomous software agents and live threads that combine people with several AI providers.
Slack puts coding agents inside project channels
Slack announced Slack Code on 20 August 2026. The product allows AI coding agents to work in project channels alongside human teammates, giving teams a shared place to direct the work and review what happens next.
Slack Code lets users create code channels, follow an agent’s activity, review proposed changes, and view live HTML previews. People can also tell an agent to change course, pause it, redirect it, or stop it. Slack says each channel maps to a particular project, while finished channels can archive themselves and leave behind a searchable record.
That setup is aimed at more than engineers. Slack says product managers, designers, and other non-engineering staff can take part without opening a terminal or learning a new tool. The company describes the result with a blunt promise: “No ticket, no meeting, no waiting – just a fix, shipped.”
Human control remains part of the design. Slack says higher-stakes actions, such as pushing code to production, require expert approval. Agents also inherit existing permissions, security policies, and administrative controls, which keeps their access tied to the rules already used by the organization.
Slack says much of today’s AI-assisted coding happens out of sight, away from the team’s normal conversations. Code channels bring that process into Slack, where people can see the work, respond to changes, and keep the project history together. Slack mentions integrations with Anthropic, Cognition, GitHub, OpenAI’s ChatGPT, and Vercel, along with agents such as Claude, Devin, Copilot, and ChatGPT. APIs will eventually open to the broader developer community.
DeepSeek builds a framework for autonomous agents
DeepSeek launched Harness on 14 August 2026. It is a software framework for autonomous AI agents that can operate external software, run code, and complete complex jobs.
Harness includes four operational modes: standard, code-focused, creative, and minimal. Those options give users different ways to direct an agent, depending on whether the task centers on ordinary work, software development, creative output, or a smaller operating setup.
DeepSeek hired Cui Tianyu, a former Jane Street engineer, in March to join its harness group. The launch places Harness alongside the broader move toward agents that do more than answer prompts. These systems can take actions in software and work through multi-step jobs, rather than stopping after producing text.
SuperApp turns the live thread into the workspace
SuperApp, Inc., formerly Instabase, launched SuperApp on 17 August 2026. The company calls it an AI collaboration super app, and it is available on the web, iOS, Android, and desktop.
SuperApp brings teams and leading AI models from Anthropic, OpenAI, Google, xAI, and others into one live thread. It combines features similar to Slack, AI models, and Google Docs, allowing people and AI to share the same conversation while keeping context and feedback in place.
“The conversation should not end when the work begins. The conversation should become the work,” SuperApp says. The product was built around that idea, with messages, source files, model contributions, tool calls, results, revisions, and approvals connected in a durable record.
Work can move across providers without moving the work itself or rebuilding its context. SuperApp is multi-provider and provider-neutral, so organizations can change models without losing their conversations or unfinished work. The organization owns and governs the record of requests, model contributions, source, tool calls, results, revisions, and approvals.
Anant Bhardwaj, CEO of SuperApp, describes the shift this way: “The prompt box was a breakthrough. It made AI personal. SuperApp makes it collaborative. Work is multiplayer. AI should be too.”
The company has also used SuperApp for its own workforce, moving employees off Slack and migrating more than 20 million historical messages. Before its public launch, the product had more than 1,000 weekly active users, with roughly 90% outside the company.
Sampriti Bhattacharyya, CEO of Navier, described using the product in terms of channels, carried-forward history, and shared documents: “The answers are genuinely thoughtful and rigorous — not fluffy. I organize my work in channels, carry history forward, and collaborate with my team on documents without losing the thread. So much of what I need is finally in one place.”
SuperApp is available now through the web, Apple App Store, Google Play, and desktop. It offers free access, with paid options for higher usage and enterprise deployments. The company is backed by Andreessen Horowitz (a16z), Greylock Partners, Index Ventures, New Enterprise Associates (NEA), Spark Capital, Tribe Capital, and QIA.
Taken together, these launches show three ways AI agents can enter team workflows: Slack places them in project channels, DeepSeek gives them a framework for operating software, and SuperApp makes the conversation itself a shared work record. The tools differ, but each treats AI as part of the working group rather than a separate box waiting for a prompt.
Based on
- Slack Code taps into collective vibe, puts AI agents into the group chat — theregister.com
- Launch of DeepSeek’s Harness marks its strategic pivot towards autonomous agentic AI | South China Morning Post — scmp.com
- Instabase Becomes SuperApp, Launches the AI Collaboration Super App | AFP.com — afp.com
