Gumloop
Tag @Gumloop in a Slack channel or CC an agent on an email thread — no separate dashboard required. Backed by a $50M Series B from Benchmark in March 2026, with Shopify, Ramp, Instacart, and Gusto among its customers.
What is Gumloop?
Gumloop, founded in 2023 by Max Brodeur-Urbas and Rahul Behal as a Y Combinator Winter 2024 company, takes a genuinely distinctive approach to where automation lives: rather than forcing users into another dashboard or admin panel, agents are embedded directly into the communication tools teams already use daily — Slack, Microsoft Teams, and email. Tag @Gumloop in a channel, CC an agent on an email thread, or trigger a workflow from a Teams message, and the agent reads the full surrounding context, formulates a plan, and executes across connected tools without anyone needing to open a separate interface. Underneath that, Gumloop combines a visual, node-based workflow builder — drag-and-drop triggers, logic steps, integrations, and AI actions onto a canvas — with genuine autonomous agent capabilities including subagent delegation and a skill-based architecture, connecting to more than 100 apps and 50+ MCP servers including Salesforce, HubSpot, and Notion. A built-in assistant nicknamed Gummie accepts natural-language instructions to build or troubleshoot workflows directly, explaining its reasoning in real time as it creates or edits nodes.
The company's growth reflects real, credible market traction: a $50 million Series B led by Benchmark closed in March 2026, and its customer roster includes genuinely well-known names — Shopify, Ramp, Instacart, Samsara, Webflow, Gusto, and Albert among them. Concrete use cases reported directly by these kinds of teams: revenue teams deploying agents that qualify leads, personalize outreach, schedule follow-ups, and generate pre-meeting briefs entirely inside Slack without leaving the sales workflow; support teams building agents that triage incoming tickets, draft context-aware responses, escalate complex cases to humans with full thread summaries, and follow up on resolved tickets to measure satisfaction. The genuinely useful, specific thing worth understanding about how credits actually get consumed: more than half of Gumloop's native nodes — text manipulation, filters, and integrations with apps like Google Sheets, Slack, and Salesforce — cost zero credits at all. AI nodes carry real costs depending on model tier: 2 credits for standard models, 20 credits for advanced models like GPT-4.1 or Claude Sonnet 3.7, and 30 credits for expert-tier models, while data enrichment and web-scraping nodes run notably higher — 60 credits to enrich a single contact, 10 credits for advanced web scraping — meaning workflows built mostly from free logic nodes stay genuinely cheap, while research- or enrichment-heavy workflows can consume credits fast.
The Slack/Teams/email embedding approach, funding round, and customer roster are drawn directly from a detailed independent 2026 review (TechTich). The specific credit-cost breakdown by node type is drawn from an independent hands-on analysis (Willo). The honest scaling trade-offs are drawn from a separate independent review (Lindy's own comparative blog).
Key features
Embedded in Slack, Teams & email
Trigger and interact with agents directly inside the tools you already use.
Visual node-based builder
Drag-and-drop triggers, logic, integrations, and AI actions onto a canvas.
Gummie AI assistant
Builds and troubleshoots workflows from natural language, explaining its reasoning.
Batch operations
Process lists of files, URLs, or records simultaneously for document work at scale.
Interfaces
Publish an input form on top of a workflow so end users never see the automation underneath.
Enterprise controls
SOC 2 Type II, VPC deployment, audit logs, and role-based access control.
Pricing
Free
- Unlimited agents and workflows, 2 concurrent runs
- Genuinely usable for small teams experimenting with AI agents
- 1 active trigger, 5 concurrent agent interactions
Paid tiers
- Text/filter/integration nodes remain free even on paid tiers
- AI nodes: 2 credits (standard), 20 (advanced), 30 (expert models)
- Enrichment and scraping: 60 credits to enrich a contact, 10 for advanced scraping
Enterprise
- Best for larger organizations with compliance and security requirements
- Dedicated support and higher concurrency limits
- Confirm current credit allowances directly given frequent updates
Workflows built mostly from free logic nodes stay genuinely cheap; research- or enrichment-heavy workflows using web scraping and contact enrichment consume credits significantly faster. Prices reflect Gumloop's published pricing as of July 2026.
Available models
Integrations & platforms
Pros, cons & best for
Pros
- Genuinely distinctive embedding directly into Slack, Teams, and email, no separate dashboard
- Zero-credit logic nodes keep simple workflows genuinely affordable
- Real, credible customer base and funding signal reflecting market traction
Cons
- Real learning curve for building complex, distributed multi-agent workflows
- Enrichment and scraping nodes can consume credits fast for research-heavy use cases
- Debugging multi-agent logic inside a visual editor can be genuinely difficult
Best for
- Revenue and support teams wanting agents embedded directly in Slack or Teams
- Operations teams batch-processing documents, CRM records, or web data
- Not the pick for teams wanting the absolute simplest, lowest-learning-curve tool
Take a look inside
Alternatives
For broader integration coverage or a similar AI-agent approach instead:
Our verdict
Gumloop's genuine differentiator is where the product actually lives — embedding agents directly into Slack, Teams, and email removes the real friction of yet another dashboard, and its credible customer roster (Shopify, Ramp, Instacart, Gusto) alongside a $50 million Series B from Benchmark reflects real, current market validation rather than hype alone. The credit system is genuinely thoughtfully designed too: more than half of native logic nodes cost nothing at all, keeping simple automations affordable, with real cost concentrated specifically in AI reasoning and data enrichment, which is a fair, transparent way to price a genuinely AI-native platform. The honest trade-off worth planning around: complex, distributed multi-agent workflows carry a real learning curve, and debugging agent logic inside a visual editor can be genuinely difficult when something goes wrong. For revenue, support, and operations teams wanting agents embedded directly in their existing communication tools, Gumloop remains a genuinely strong, well-funded, distinctive choice.
FAQ
How do I actually use Gumloop's agents day-to-day?
By tagging @Gumloop directly in a Slack channel, CCing an agent on an email thread, or triggering a workflow from a Teams message — no separate dashboard is required, and the agent reads the surrounding context before acting.
Which parts of a Gumloop workflow actually cost credits?
Text manipulation, filters, and standard app integrations (Google Sheets, Slack, Salesforce) cost zero credits — real cost comes from AI nodes (2-30 credits depending on model tier) and data enrichment or web-scraping nodes (up to 60 credits to enrich a single contact).
Who uses Gumloop?
A genuinely credible customer roster including Shopify, Ramp, Instacart, Samsara, Webflow, Gusto, and Albert, following a $50 million Series B led by Benchmark that closed in March 2026.
What is Gummie?
Gumloop's built-in AI assistant that accepts natural-language instructions to build or troubleshoot workflows directly, creating or editing nodes while explaining its reasoning in real time.
Is Gumloop's free plan good enough for real use?
For small teams experimenting with AI agents, yes — it includes unlimited agents and workflows with a real, usable credit allowance, though ongoing production use at volume typically requires a paid tier.
Is Gumloop hard to learn?
Simple workflows using free logic nodes are genuinely approachable, but complex, distributed multi-agent workflows carry a real learning curve, and debugging agent logic inside the visual editor can be difficult when something misbehaves.