Gumloop

What is Gumloop?

Gumloop is a web-based AI agent builder for companies, pitched as "the multiplayer AI agent builder. Let anyone at your company build agents with any AI model and any integration, while IT controls access and spend." An employee connects the apps a task touches, and colleagues reuse the resulting agent in chat, Slack, Teams or email.

The service is operated by AgentHub Inc., which does business as Gumloop. Independent reporting from March 2026 says Max Brodeur-Urbas co-founded Gumloop in mid-2023 to help non-technical employees automate repetitive tasks with AI. The same report covers a $50 million Series B led by Benchmark. Its accelerator profile lists the Winter 2024 batch and a San Francisco base.

The product has shifted over time. Its earlier building block, Flows, was renamed Workflows, and the platform now leads with agents that call tools, run on triggers and improve their own skills. Workflows still exist and are covered below, but the pricing page marks them as a legacy feature.

Core features

Agents and subagents

• Multi-agent orchestration: teams can "Orchestrate complex processes with multiple agents inside deterministic workflows. Agents can even call on subagents."
• Built-in quality checks: "Built-in evals let you measure quality, catch regressions, and ship improvements with confidence," and agents can reflect on past runs to tune themselves.

Triggers written in plain English

Triggers let an agent run without anyone typing a prompt. Beyond schedules and app events, "Agents create and manage their own automations during conversations," such as "remind me to check emails every weekday at 9am". A single trigger can also poll several services, such as Gmail, Salesforce and Stripe, in one check.

Skills, Company Brain and artifacts

• Skills: reusable playbooks. "Agents can create, edit, and improve their own skills at runtime, so they get better and more efficient." Skill folders can also be synced from a GitHub repository.
• Company Brain: a shared knowledge index built from sources such as Notion, Drive, Slack, GitHub and Confluence. It is meant to "Connect your team’s shared knowledge into a centralized, always up-to-date brain agents and humans can use."
• Artifacts: agents can produce reports, decks and small apps. "With just a prompt, build shareable files and web apps that pull live data from your connected tools."

Model choice

Gumloop is model-agnostic: the pricing table lists 35+ models, with "35+ and custom proxy support" on Enterprise. Besides the Auto router and presets, you can "browse by provider: Anthropic, OpenAI, Google, DeepSeek, MiniMax, Z.ai, and more." The documentation says new models are usually available in Gumloop within a day of their public release.

Sharing and roles

Access can be granted as "owner, editor, viewer, or use-only access to one person, a team, or your whole organization at once," so a builder can let colleagues chat with an agent without exposing its configuration.

Workflows (legacy)

Workflows are the older node-based automation builder, the place "where you can orchestrate agents and deterministic steps," in the words of the changelog note that renamed Flows to Workflows. The pricing table now labels the row "Workflows (Legacy)". The changelog still ships workflow changes: the 10.22.0 release on August 20, 2026 added Monday.com board activity as a workflow trigger. Gumloop has not published an end-of-life date for workflows in the pages checked for this review.

Guide

Two documented account tasks affect cost more than any feature choice.

Add your own model key (BYOK)

1. Open Connectors in your personal settings (Standard plans) or Organization → API keys (Enterprise, where org keys override personal and team keys).
2. Click Add Key, paste an OpenAI, Anthropic, Google AI, Perplexity, SpaceXAI (xAI) or Fireworks key, and save.
3. Run a short agent chat on a model that key covers.
4. Click the credit count in the chat header; the model portion of that call should show 0 credits.

Cancel a trial or plan

1. Go to Settings → Organization → Usage & Limits.
2. In the Plan card, select Manage Subscription. This opens the Stripe billing portal in a new tab.
3. Choose Cancel plan and confirm.

In an organization, only the Admin role can open the billing portal. Enterprise contracts are invoiced separately and changed through the accounting team rather than this portal.

Gumloop use cases and examples

Gumloop's showcased agents cluster around go-to-market, support and internal data work:

• Data Analysis Agent: "A reasoning agent that answers questions from your data warehouse."
• Support Agent: "An agent that triages bugs, creates tickets, and spots support patterns automatically."
• Meeting Prep Agent: "An agent that automatically briefs you before every meeting with context from across your tools."
• Call Analysis Agent: "An agent that analyzes call recordings to surface objection patterns, coaching insights, and competitive intelligence."

Monitoring is the other recurring pattern: "Agents can detect percentage drops, time-without-reply, cross-system mismatches, or any other condition you can think of."

The strongest outcome figure on the site is a customer quote, not an independent measurement. Gusto's head of insights and operations is quoted as saying: "The partnership review agent cuts prep time from 30–45 minutes per customer to under five minutes, and it results in a 31% higher win rate and 44% larger deals."

Who is it for

Gumloop targets companies that want business teams, not only engineers, to build no-code AI agents. In Gumloop's words, "Technical and non-technical teams alike can collaborate on AI agents together, and deploy them where the team already works." Press coverage of the Series B relays the company's claim that teams at Shopify, Ramp, Gusto, Samsara, Instacart and Opendoor use it; that customer list comes from Gumloop rather than from independent verification.

• Good fit: sales, support, data and operations teams that want one person's agent to be reused across the organization.
• IT and security leads: the company states that "Every observability or access control feature an IT leader could ask for is available on our enterprise tier." Admins can also write guardrails that "block, tag, or log actions before they run."
• Enterprises standardizing MCP: Gumstack, launched in March 2026, is "our security, observability, and MCP infrastructure product for enterprise customers."
• Less suited: individuals looking for a permanently free tier, since the visible plans are Pro and Enterprise only.

Platforms

Gumloop runs in the browser, and agents can also be reached from other tools.

• Workplace chat and email: "Interact with agents like co-workers in Slack, Teams, and email. Simply tag @Gumloop and ask." WhatsApp is listed as coming soon.
• Chrome extension: browser recordings let you "Record your browser actions and replay them with a single click. No coding required," and the extension can scrape the page you are viewing into Gumloop.
• CLI: "Sign in once, then run agents, call MCP tools, search Company Brain, and manage skills and artifacts from your terminal."
• API and SDKs: a REST API with webhooks, plus the option to "Build on Gumloop from your own code with the Python and JavaScript SDKs."
• Self-hosted network option: Enterprise customers can "Deploy Gumloop inside your own AWS, Azure, or Google Cloud environment to keep data in your network."

Integration counts are vendor figures and differ by page (as captured on September 28, 2026): the homepage cites 300+ Connectors. The integrations page says "Connect any AI agent to 250+ MCP servers, zero setup."

Pricing

Gumloop pricing has two published tiers plus usage-based credits. Figures below were captured on September 28, 2026.

The Pro card reads "Starts at $37 / month" and offers a 14-day free trial. Its 20,000 monthly credits are "7,400 credits at the $0.005 list price, plus 12,600 bonus credits."

How credits are charged

One credit equals $0.005, so "$1 buys 200 credits." On top of model token costs, each agent chat is billed for:

• Tool calls: "Every successful tool call costs at least 1 credit. Failed calls are not charged." Paid data tools such as enrichment or scraping add their own charge on top.
• Compute: "5 credits per session-minute of active processing, with a minimum of 1 credit per response."
• Orchestration fee: 8% of the three items above. Gumloop's worked example: "A chat that spent 400 credits on Chat & Reasoning, 10 on Compute, and 15 on Tool Calls pays a 34 credit fee (8% of 425, rounded up), for 459 credits in total." Enterprise plans can negotiate a different orchestration fee.

Bringing your own key

With BYOK, "Model calls that the key can serve cost 0 credits," but the fee changes: "Agent chats on BYOK use a 16% orchestration fee (instead of 8%) on the run’s value." It "Requires Pro plan or higher and your own OpenAI, Anthropic, Google AI, Perplexity, SpaceXAI, or Fireworks AI account."

Overage, rollover and renewals

• Rollover: "Credits don’t roll over month to month, except on Enterprise plans."
• Overage: you must "Enable credit overage on your Usage & Limits page to keep running past your monthly credits, billed at $0.005 per credit." "Pro overage is always capped. The default ceiling is 1,000,000 overage credits per billing period ($5,000), and you can set a lower cap on that page."
• Trial: "The trial requires a card, is a one-time offer per customer, and rolls into a paid Pro subscription when it ends unless you cancel first."
• Renewal: the terms state that "your Subscription will automatically renew under the exact same conditions unless you cancel it."
• Cancellation timing: "Cancelling is never immediate: you keep access until the end of the current billing period or trial, and the Plan card shows a Cancels badge with that date."
• Refunds: "Except when required by law, paid Subscription fees are non-refundable."

The pricing page's structured data is not fully aligned with the visible table. It lists a "Free plan" offer at $0 described as "Free 14-day trial available", while the visible pricing table shows only Pro and Enterprise, and the documentation describes the free period as a trial of Pro rather than a free plan.

Gumloop alternatives

Gumloop competes with both automation platforms and newer agent builders. Independent reporting on the Series B named its main rivals: "The startup faces stiff competition from established automation platforms like Zapier and n8n, as well as specialized agent builders like Dust." The same report added that "Anthropic’s Claude Cowork allows users to create autonomous agents without writing a single line of code."

The investor quoted in that coverage argues that "Gumloop’s model-agnostic approach is precisely what will keep attracting customers," which reflects Benchmark's view rather than a product test. Gumloop itself publishes comparison pages against ChatGPT Agent, Claude Cowork, Dust, Glean, Grok Bot, n8n, Relevance AI and Zapier; those are vendor-written and not a neutral benchmark.

Limitations

Cost predictability

• Variable charges: "The same agent might cost 2 credits for a quick question and 200 for a deep research task."
• Spend visibility is tiered: the org-wide Insights dashboard is an Enterprise feature; Pro users rely on the usage tables under Settings.
• BYOK coverage is partial: "A key only covers models that provider can serve," so mixing models can still consume credits.

Operational limits

• Concurrency: Pro allows 25 concurrent agent interactions and Enterprise 100 by default, counted across the whole organization, from the web UI, Slack, Teams, the API and triggers. When Pro hits its cap, the documentation states "At limit + Pro — request is rejected", whereas Enterprise requests are queued.
• Trigger auto-disable: "Gumloop pauses a trigger automatically after 3 consecutive failed runs." Expired credentials and exhausted credits are the most common causes.
• Paid plan required for schedules: "Recurring triggers stop running when the owning account no longer has an active paid plan."
• Overage cap stops agents: once the configured cap is hit, agents stop rather than billing further.

Privacy, training and rights

• Model training: "For premium users, we do not use any of your uploaded data, flows, or chats with agents that you create to train any AI whatsoever." The policy frames this commitment around premium users; it does not spell out the same wording for other accounts.
• Model providers: "We have an agreement directly with OpenAI and Anthropic to ensure they’re also committed to not training on any data sent to them via Gumloop’s API." The policy names no such agreement for other model providers.
• Google data: Gumloop states "we do not use Google Workspace APIs to develop, improve, or train generalized AI and/or ML models."
• Compliance claims: the homepage states "SOC 2 Type II. Independently audited and compliant." and "GDPR. Zero Data Retention agreements for third-party models." These are vendor statements; no audit report was reviewed.
• Your content: the terms say "You retain any and all of your rights to any Content you submit, post or display on or through Service and you are responsible for protecting those rights."
• Document dates: the privacy policy shows "Effective date: 06/26/2026". The terms of service show "Last updated: 02/07/2025".
• Age: "Service is intended only for access and use by individuals at least eighteen (18) years old."

No independent reviews could be checked for this page: the main review platforms blocked both direct and rendered access on September 28, 2026, so the limitations above come from Gumloop's own documentation and terms.

FAQ

Q1. Is there a free Gumloop plan?

The visible pricing page offers Pro and Enterprise only. New accounts get a 14-day Pro trial that needs a card and converts to paid Pro unless cancelled.

Q2. What happens to my agents if I cancel?

Plan access ends on the cancellation date and agents stop running, but "Your agents, workflows, skills, Brain sources, chats, and other workspace data stay in your account when you cancel." You can resubscribe later.

Q3. Can I still build workflows?

Yes. Workflows remain available and receive updates, but the pricing table labels them "Workflows (Legacy)" and the product now centers on agents.

Q4. Who pays when colleagues use a shared agent?

Usage inside an organization draws on one shared credit pool, whoever starts the chat. "If you share an agent with someone outside your organization, their usage bills to their account or org, not yours."