Gumloop
Gumloop is a no-code platform where anyone at a company can build, share and govern AI agents. It connects hundreds of business apps, deploys agents inside Slack, Microsoft Teams and Gmail, and leaves IT in control of access.
What is Gumloop?
Gumloop is an AI agent platform published by AgentHub Inc. Its premise is blunt: understanding a task is the only prerequisite to automating it, so the people who already know a process should be the ones who automate it, with no learning curve and no code.
The product began as a no-code, drag-and-drop workflow builder. The March 2026 Series B announcement marks its pivot to agents, and the company openly invites anyone still holding the old picture to look again. Three pieces make up the offer today. Gumloop Agents are proactive agents that anyone, from a marketing intern to the CEO, can assemble in minutes, wire to hundreds of applications and deploy natively in Slack, Microsoft Teams and email. Gumloop is the platform where teams collaborate on, build, share and orchestrate those agents. Gumstack targets security teams and tracks data use across an organisation's agents, including tool calls made outside Gumloop from Claude Code, ChatGPT, Cursor or in-house agents.
Two mechanisms give agents their context. Company Brain centralises company knowledge and keeps it current, syncing from any connector while preserving each source's original permissions. Skills let agents write their own playbooks as SKILL.md files, improve themselves across runs and execute their own code in a sandbox; a GitHub repository can feed an organisation's skill library directly.
The platform is deliberately model-agnostic: 35+ models on the paid tier, custom proxy support for enterprises, bring-your-own API keys, automatic fallbacks and a stated preference for open-source models to cut per-task cost. Agents run an Execute, Reflect and Learn loop, with built-in evals to measure quality and catch regressions.
The governance layer carries the enterprise pitch. Administrators get real-time credit and spend monitoring, budgets and approvals, audit logging, role-based access control, SAML SSO and SCIM, per-role model restrictions, scoped credential management, plain-English guardrails that block, tag or log an action before it runs, per-tool MCP tracing, and optional VPC deployment inside the customer's own AWS, Azure or Google Cloud. Developers get a documented API, a CLI and MCP server hosting.
What it does
- Build an AI agent without code and connect it to hundreds of business applications
- Deploy agents where the team already works: mention them in Slack channels, Microsoft Teams chats or let them draft and triage in Gmail
- Connect company knowledge into a single searchable Company Brain that respects each source's original permissions
- Share an agent with one person, a team or the whole organisation at owner, editor, viewer or use-only level
- Trigger agents on application events or on a recurring schedule, and let them produce shareable dashboards, decks and documents
- Cap and monitor AI spend by agent, team or organisation, with approval flows for costly actions
- Trace every MCP tool call through one logging and analytics layer, including calls made outside Gumloop
When to use Gumloop / When not to
A quick filter to help you decide if Gumloop is the right fit.
When to use Gumloop
- Business teams who understand a manual process and want to automate it themselves, without writing code or waiting on engineering
- Go-to-market teams in sales, marketing and support looking for ready-made agent patterns such as CRM upkeep, lead qualification, outbound research, ticket triage and campaign reporting
- IT and security leaders who need AI adoption to come with SSO, SCIM, role-based access, audit logs, model restrictions and spend caps built in
- Operations and data teams that want live answers pulled from company knowledge and connected tools rather than another dashboard to maintain
- Larger organisations with residency or network constraints, who can deploy Gumloop inside their own AWS, Azure or Google Cloud environment
When not to use Gumloop
- Individuals and hobbyists on a tight budget: there is no permanent free plan, and the entry point is a paid subscription once the 14-day trial ends
- Anyone who needs the widest possible catalogue of app connectors, since Gumloop's own comparison puts it at 100+ connected apps against 8,000+ for a generalist competitor
- Teams that work primarily from phones, as no iOS or Android application is offered
- Non-English-speaking teams who need a localised interface, because no other interface or processing language is advertised
- Buyers who need a fixed, predictable monthly bill, since credit consumption and an 8% orchestration fee sit on top of the headline price
How to use Gumloop
A typical end-to-end flow, from setup to results.
- Start from the Get Started button on the homepage, or book a demo through the contact form if you are evaluating the enterprise tier
- Open the 14-day free trial of the Pro plan to work with the product before committing
- Describe the task you want automated: agents are assembled conversationally, so the prerequisite is knowing the process, not the tooling
- Attach the connectors the agent needs, choosing from the catalogue of business applications and MCP servers
- Point the agent at your company knowledge by setting up Company Brain sources, using the short configuration chat to pick exactly what is synced
- Give the agent its instructions and skills, and let it write and refine its own playbooks as it runs
- Choose how the agent starts: an application event, a recurring schedule, or a direct message from a colleague
- Share it at the right level, owner, editor, viewer or use-only, with one person, a team or the whole organisation
- Push it into daily work by mentioning it in a Slack channel, bringing it into Microsoft Teams, or letting it draft in Gmail
- Watch adoption, credit consumption and outcomes in Insights, set spend caps, and go deeper through the API, CLI or MCP server when needed
Pros & Cons
Pros
- Genuinely reachable by non-technical staff, a point echoed in published testimonials from Instacart, Webflow and Albert
- Enterprise governance comes as standard rather than as an upsell: SOC 2 Type II, SSO, SCIM, role-based access, audit logs, spend caps and MCP tracing
- Model-agnostic by design, with 35+ models, personal API keys, automatic fallbacks, proxy routing and an open-source push the site prices at $0.42 down to $0.071 per task
- Agents live where the work already happens, in Slack, Teams and Gmail, instead of asking teams to adopt yet another interface
- A single credit currency spans agents, workflows and integrations, and the subscription bundles access to current models and to enrichment, research and scraping tools
- Observability reaches beyond the platform itself: Gumstack logs tool calls from Claude Code, ChatGPT, Cursor and in-house agents too
- Unlimited seats and teams from the entry-level plan, backed by a substantial learning ecosystem of university courses, cohorts, workshops, a forum, templates and vetted experts
Cons
- No permanent free plan: once the 14-day trial ends, the entry point is a paid subscription at $37 per month
- Total cost is hard to forecast, since credit consumption and an 8% orchestration fee sit on top of the headline price, and enterprise pricing is not published
- No postal address appears anywhere on the site, on any of the twelve pages collected
- No subprocessor list is published and no DPA is advertised; the trust centre returns nothing without JavaScript, so due diligence depends on a sales conversation
- The no-training commitment in the privacy policy is worded for premium users and says nothing about other accounts
- Fewer connected applications than the generalist competition, at 100+ against 8,000+ for Zapier, using Gumloop's own published figures
- No mobile application, no advertised interface language other than English, and several governance features, VPC among them, are reserved for the enterprise tier
Pricing & Plans
There is no permanent free plan. The lowest paid entry point is the Pro plan, advertised as starting at USD 37 per month and preceded by a 14-day free trial. Pro includes 20,000 credits per month and carries an 8% orchestration fee on top; the Enterprise plan is priced on request, with custom credit allowances and discounts available on the orchestration fee. Billing runs monthly or annually depending on the plan chosen, renews automatically, and subscription fees are non-refundable except where the law requires otherwise.
- unlimited agents
- seats and teams
- 35+ models
- bring-your-own API keys
- 20
- 000 credits per month
- 8% orchestration fee
- Company Brain
- everything in Pro plus custom model proxy support
- custom credit allowances
- discounts on the orchestration fee
- dedicated Slack support
- optional embedded Gumloop expert
- optional virtual private cloud deployment
- organisation-wide connector policies and guardrails
- custom MCP server hosting
Data, GDPR & hosting
A consolidated view of how Gumloop handles your data.
GDPR overview
GDPR is addressed concretely rather than in passing. A dedicated section sets out the rights of EU and EEA residents, access, rectification, erasure, restriction, objection, portability and withdrawal of consent, with the right to complain to a supervisory authority; equivalent UK GDPR rights are recognised. Gumloop is certified under the EU-U.S. Data Privacy Framework and its UK Extension with the U.S. Department of Commerce, commits to cooperate with EU data protection authorities and the ICO, accepts binding arbitration and remains liable for onward transfers, under FTC jurisdiction. A Data Protection Officer is appointed and reachable at privacy@gumloop.com. No Article 27 EU representative is named anywhere on the site. The privacy policy in force is dated 26 June 2026, and the site also advertises SOC 2 Type II certification.
Who owns the data?
AgentHub Inc., trading as Gumloop, acts as data controller for the personal data it collects through its site. The privacy policy states plainly that the company neither sells nor rents personal information to third parties for any purpose, and that the user is the sole owner of their personal data, free to request disclosure or deletion at any time. Inside a workspace, control sits with the organisation's administrator: they restrict access to organisations, workspaces and individual flows, revoke it, and can have a member's profile and data deleted on request. Third-party credentials are walled off, so no user can reach another user's stored secrets.
Reuse rights
Gumloop lists its purposes explicitly: running and maintaining the service, customer support, analysis and improvement, usage monitoring, fraud and abuse detection, billing, account notices, and marketing of similar goods and services unless the user opts out. Data obtained through Google Workspace APIs is ring-fenced and is not used to build or train generalised AI models, nor passed to third-party AI tools for that purpose. On model training the policy commits, for premium users, that no uploaded data, no flow and no agent conversation is used to train any AI, and states that direct agreements with OpenAI and Anthropic extend that commitment to data sent through Gumloop's API. Zero Data Retention, marketed as Incognito mode, covers sensitive runs. Users keep and reuse their own content and outputs freely within the service; analytics run through Google Analytics and Firebase, payments through Stripe, which never exposes card details to Gumloop.
Data retention & training
Hosting summary
Data is hosted and processed in the United States. The privacy policy states that information from users located outside the United States is transferred there and processed there, and the transfer basis claimed is certification under the EU-U.S. Data Privacy Framework and its UK Extension with the U.S. Department of Commerce. No data centre, cloud provider or specific hosting region is named for the standard service, and no EU data residency option is advertised. Enterprise customers can take a different route: Gumloop offers optional deployment inside the customer's own AWS, Azure or Google Cloud environment, keeping data within their network. On operational security the policy describes nightly database backups, weekly manual penetration testing against its own product, workspace access restricted by default, static code analysis through CASA, and strict isolation of third-party credentials so no user can reach another's. Internally, access is restricted by default, devices are wiped when employment ends, and two-factor authentication is enforced on sensitive accounts.
Things to keep in mind
Risks and trade-offs to weigh before adopting Gumloop.
- The no-training promise is scoped: the privacy policy commits only for premium users and is silent on what happens to everyone else's data
- No postal address is published anywhere on the site, so the only verifiable territorial anchors are Delaware governing law and data processing in the United States
- No subprocessor list and no DPA are publicly available, and the trust centre serves no content without JavaScript, which pushes documentary due diligence into a sales conversation
- The 8% orchestration fee is charged on top of the subscription and credit costs, so the advertised $37 per month is not the total cost of ownership
- Legacy labelling on the original workflow builder leaves its long-term future open, which matters if you are migrating existing automations onto it
- Agents that write their own code, self-improve and act across connected systems shift real authority away from human review, so guardrails, approvals and audit logs deserve configuring before rollout, not after
- Handing routine judgement to agents can quietly erode a team's own grasp of the process; the volume figures on the homepage render as zero without JavaScript and should not be taken as verified either
Setup & Integrations
Technical difficulty
Low for everyday use, moderate to high for the administrative and developer layers. Gumloop claims no learning curve: agents are assembled conversationally, connectors are ready to use, and MCP servers are advertised as zero setup. Real effort sits with the administrator, who configures credentials, roles, permissions and Company Brain sources, and with identity teams handling SAML SSO and SCIM. Advanced work through the API, CLI, custom MCP integrations, managed tunnels or VPC deployment needs engineering skills. Support is substantial: documentation, university courses, week-long cohorts, workshops, webinars, a forum, templates, vetted experts, and dedicated Slack support on the enterprise tier.
Deployment
Integrations
Behind Gumloop
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Gumloop.
Frequently asked questions
How much does Gumloop cost?
Is there a free plan or a free trial?
Do I need to know how to code?
Which AI models can I use?
Where can agents be used day to day?
Is my data used to train AI models?
Is there an API?
What security certifications does Gumloop hold?
Can Gumloop run inside our own cloud?
Is there a minimum age to use Gumloop?
Should you pick Gumloop?
Gumloop makes a coherent bet: put agent building in the hands of the people who already understand the work, then wrap it in the controls an IT department will insist on before letting that happen. The execution matches the pitch. Agents reach staff inside Slack, Teams and Gmail rather than behind another login; Company Brain and Skills give them real context; and the governance layer, with SOC 2 Type II, SSO, SCIM, role-based access, audit logs, guardrails, spend caps and optional VPC deployment, is unusually complete for a product sold to non-technical users. A USD 50 million Series B led by Benchmark in March 2026, named customers such as Shopify, Instacart, Gusto and Samsara, and a changelog that shipped roughly twenty entries in six weeks all point to a product with momentum behind it.
The reservations are mostly commercial and documentary. The permanent free plan is gone, so evaluation now runs on a 14-day trial before a USD 37 monthly commitment, and the real bill depends on credit consumption plus an 8% orchestration fee that the headline price does not show. Enterprise pricing is quoted rather than published. On paper trail, the site is thinner than its security posture suggests: no postal address anywhere, no published subprocessor list, no advertised DPA, and a trust centre that renders nothing without JavaScript. The no-training commitment is worded for premium users and stops there.
For an organisation that wants to spread AI across business teams while keeping IT in the loop, Gumloop is a serious candidate and worth a trial. For a solo user, a small budget or a procurement team that needs documentation before a sales call, it is a harder fit. Ask for the DPA and subprocessor list early.
- Choosing a selection results in a full page refresh.
- Opens in a new window.