Kollab

Overview

Kollab is an AI-native workspace where teammates and AI agents share the same project surface — tasks, documents, and chat all sit alongside the agents that act on them. Instead of treating AI as a chat sidebar, Kollab lets agents pick up work, run multi-step Skills, and post results back into the channels and tools your team already uses.

Kollab is newly launched and publicly available to try. It targets cross-functional teams that mix engineers, operators, and PMs — groups that lose hours rewriting context for every new tool or new agent. By keeping a persistent memory of decisions, project state, and reusable workflows, Kollab tries to make agent output less of a one-off prompt result and more of an organizational asset.

If you've been stitching together Slack, Notion, Linear, and a half-dozen AI assistants, Kollab's bet is that the workspace itself — not the model — is what should be shared.

Key Features

• Shared AI workspace — A single project surface where humans and agents see the same tasks, files, and progress, so handoffs don't lose context between people, tools, or sessions.

• Agent Skills — Reusable, multi-step routines anyone on the team can install by name. Skills capture how a workflow should be run, so the second person doesn't re-explain the playbook to the model.

• Bots inside Slack, Feishu, Discord, and Telegram — Agents live where the team already chats. Trigger an agent in Slack, and the result syncs back into the Kollab workspace with full provenance.

• Tool Connectors — Pre-built integrations for Notion, Linear, Figma, Google Drive, GitHub, Gmail, and Canva, plus MCP, API/HTTP, and app-auth connector support, letting agents read and write through authorized connector access across your existing stack.

• Knowledge base with traceable citations — Documents, meeting notes, and archives become a retrievable knowledge base. When an agent answers, it links the source so reports and analyses stay auditable.

• Persistent memory — Agents retain organizational context across projects: prior decisions, product positioning, strategic direction. Functions like a built-in memory system for agents, so you skip re-uploading briefs every time a new project kicks off.

• Scheduled tasks & AgentCore — Timed agents can call MCP tools, query the knowledge base, and post to channels on a schedule. AgentCore is described by a Kollab maker as a long-running agent with its own filesystem and built-in browser access for heavier execution work; label this as an early maker-stated capability rather than a fully documented official feature.

Integration Guide

Kollab's value compounds with the number of tools you connect, so most teams set up integrations early. The integrations fall into three layers:

Communication bots (where agents live)

• Slack — Most common deployment. Add the Kollab bot, and team members can trigger Skills, ask agents questions, and pull workspace context without leaving Slack.
• Feishu / Lark — Native bot for teams operating in Asia-Pacific.
• Discord — Useful for community-led teams or open-source projects.
• Telegram — Lightweight option for small teams or async coordination.

Tool Connectors (where agents read/write data)

• Notion — Pull pages, sync meeting notes, push agent-generated drafts back into a workspace database.
• Linear — Create issues, query backlog state, route triage decisions through an agent.
• Figma — Reference design files and frames as agent context.
• GitHub — Read repositories, issues, and pull requests; useful for engineering-side Skills.
• Google Drive / Gmail — Use connected Drive or Gmail data as authorized workspace context for agent workflows; verify current in-product permissions before relying on email send automation.
• Canva — Pull design assets into agent workflows.

MCP layer (custom integrations)

• Kollab supports Model Context Protocol, API/HTTP, and app-auth style connectors, so custom internal tools, databases, or third-party MCP servers can plug into the connector layer without waiting for a first-party connector. Treat the 30–50 MCP figure as a maker-stated Product Hunt note, not a fixed official plan quota. This is what makes Kollab usable for teams with bespoke internal stacks rather than just standard SaaS combinations.

The setup pattern most teams follow: install the Slack (or Feishu/Discord) bot first → connect Notion + the company's project tracker → import existing playbooks as Skills → expose internal systems via MCP last.

Pricing & Plans

Kollab uses a credit-based FREEMIUM model. Daily refresh credits (200/day) are shared across all tiers; paid plans add a much larger monthly subscription credit pool on top. Team seats are unlimited on every plan.

Top-up credits (one-time, paid workspaces only) — 1,500 credits for $5, 3,000 for $10, 15,000 for $50. Useful for paid workspaces that need temporary extra credits without changing the subscription tier.

Notes:

• The Free plan refreshes daily up to a 2,000-credit monthly cap, then stops until next month. This is enough to evaluate the product seriously but not run sustained agent workloads.
• Pro is positioned for active small teams. The 30 concurrent sessions ceiling matters more than the credit pool for most users.
• Max is built for heavy agent users — the 80,000-credit pool and unlimited storage suggest it's targeted at teams running scheduled or long-running agents continuously.

Pricing follows Kollab's current public billing rules and may be updated prospectively under the service terms.

Best For

• Cross-functional teams (engineering + ops + PM) tired of re-explaining context to a different AI tool in every app
• Operators who already live in Slack/Feishu and want agents that act there, not in another tab
• Teams with internal workflows worth codifying — Skills pay off when the same playbook gets reused 10+ times
• Small-to-mid companies with bespoke internal systems that need MCP-style custom integrations rather than standard SaaS plumbing
• Beta-tolerant teams who want to shape an AI agent workspace early rather than wait for a polished v1

FAQ