Topolearn

Learn Gumloop

No-code AI agent platform: agents that use your apps, run on triggers and live in Slack

23 lessons · 4 hours · 4 levels · covers Gumloop as of Oct 2026 (agent platform, release 12.x)

Gumloop is a no-code platform for building AI agents that work with your company's apps and data. You give an agent instructions, connectors (Gmail, Slack, Salesforce and many more), skills and knowledge, then run it by chat, on schedules and app events, by webhook, or inside Slack, Teams and email. It started as a drag-and-drop workflow builder; since 2026 it is agent-first, and the older workflows are kept as a legacy feature.

Sign in free to track progressOfficial docs ↗

Before you start

Gumloop builds on these; the tree links to the exact lessons you need.

AI & LLM fundamentalsEssentials level

How deep do you want to go?

1. EssentialsRead and write simple code, do the everyday basics7 lessons · 58 min in total
2. PracticalUse it productively at work, with the usual idioms15 lessons · 2 hours in total
3. AdvancedThe less common features, performance and trade-offs20 lessons · 3 hours in total
4. ExpertInternals and the edge cases experts know23 lessons · 4 hours in total

The Gumloop tree

GumloopWhat Gumloop is todayWhat Gumloop istodayChatting with an agent: tasksChatting with anagent: tasksConnectors: linking your appsConnectors: linkingyour appsGumball: your personal agentGumball: yourpersonal agentCreating a custom agentCreating a customagentChoosing a model: Auto, Balanced, SmartestChoosing a model:Auto, Balanced, Sm…Approvals and human in the loopApprovals andhuman in the loopTriggers: run agents automaticallyTriggers: run agentsautomaticallyWebhook triggers and the Gumloop APIWebhook triggersand the Gumloop APISkills: reusable playbooksSkills: reusableplaybooksBrain: company knowledge for agentsBrain: companyknowledge for agentsChannels: Slack, Teams, email, hosted pagesChannels: Slack,Teams, email, host…Artifacts: files your agent createsArtifacts: files youragent createsCredits: what costs whatCredits: what costswhatLegacy workflows: workbooks, nodes, subflowsLegacy workflows:workbooks, nodes,…What a large language model isWhat a largelanguage model isCalling a model through an APICalling a modelthrough an APIWriting clear promptsWriting clear promptsTokensTokensHallucinations and other limitsHallucinations andother limits
Essentials and Practical shown (15 of 23 lessons); the full list is below. Click a lesson to read it.

All 23 lessons, in order

Gumloop basics

  1. What Gumloop is today · Essentials · 8 min

    Understand Gumloop as an AI agent platform, what an agent is in plain words, and what happened to the older workflow builder.

  2. Chatting with an agent: tasks · Essentials · 7 min

    Give an agent work in chat, attach files, steer it while it runs, and find past work on the Tasks page.

  3. Connectors: linking your apps · Essentials · 9 min

    Connect accounts like Gmail, Slack or Salesforce so agents can use them, and choose whose account an agent acts with.

  4. Gumball: your personal agent · Essentials · 7 min

    Use Gumball, the private agent every account gets, for daily catch-ups, meeting prep, inbox triage and building other agents.

  5. Legacy workflows: workbooks, nodes, subflows · Practical · 10 min

    Read and maintain Gumloop's older workflow builder: workbooks, nodes on a canvas, @-references, subflows, runs and the Chrome extension.

Building agents

  1. Creating a custom agent · Essentials · 11 min

    Build a custom agent: connectors, a real test task, instructions, abilities, and saving your changes.

  2. Choosing a model: Auto, Balanced, Smartest · Essentials · 7 min

    Pick the AI model behind an agent: the Auto router, the Balanced and Smartest presets, or a specific named model.

  3. Approvals and human in the loop · Essentials · 9 min

    Decide which actions an agent may take on its own and which need your OK, with Tool Management presets and approval cards.

Triggers and channels

  1. Triggers: run agents automatically · Practical · 10 min

    Start agents without anyone typing: schedules, one-time runs and app events such as a new email or Slack message.

  2. Webhook triggers and the Gumloop API · Practical · 10 min

    Let any system start an agent through a webhook URL, and know what the Gumloop API offers for deeper integration.

  3. Channels: Slack, Teams, email, hosted pages · Practical · 10 min

    Put agents where people already work: Slack and Teams channels, an email inbox, or a hosted chat page.

  4. Custom triggers created with AI · Advanced · 9 min

    Describe a condition to watch for across apps and let the agent build, test and run a custom polling trigger, aware of its cost.

Skills, knowledge and files

  1. Skills: reusable playbooks · Practical · 10 min

    Package a repeatable process as a skill: instructions plus optional scripts, references and templates that agents load only when needed.

  2. Brain: company knowledge for agents · Practical · 11 min

    Connect Drive, Notion, Slack and other sources to Brain so agents can search your documents and answer with citations.

  3. Artifacts: files your agent creates · Practical · 8 min

    Work with the reports, spreadsheets, images and dashboards an agent produces, including versions, sharing and password links.

Teams, quality and safety

  1. Credits: what costs what · Practical · 8 min

    Understand what uses credits (model work, tool calls, compute, Brain), how to see a task's cost, and how to spend less.

  2. Sharing, teams and roles · Advanced · 11 min

    Share agents safely: organisation and teams, Owner and User roles, team connectors, agent-owned credentials and roll-outs.

  3. Evaluations: measuring agent quality · Expert · 11 min

    Grade every completed agent conversation against your criteria, flag or notify on failures, and track quality over time.

  4. Gumloop's MCP server: use Gumloop from Claude · Expert · 10 min

    Connect AI assistants like Claude or Cursor to Gumloop's MCP server to run agents, manage skills, search Brain and start flows.

  5. Designing safe, reliable agents · Expert · 12 min

    Combine least privilege, approvals, incognito, evaluations and failure handling into agents you can trust with real work.

Advanced agent abilities

  1. Subagents: delegating work · Advanced · 10 min

    Let a coordinator agent hand parts of a job to clones of itself or to specialist agents, and combine the results.

  2. The code sandbox and Vault secrets · Advanced · 11 min

    Understand the private computer each task gets for running code and handling files, and store API keys safely in the Vault.

  3. The agent browser, logins and handoffs · Advanced · 10 min

    Let an agent use a real web browser for sites without connectors, with saved logins, live viewing and handoffs to you.