Gumloop started as a drag-and-drop workflow builder and has become an agent platform. Its pitch is that understanding a task should be the only prerequisite to automating it, and the company raised a $50 million Series B led by Benchmark in March 2026 to push that further. Teams at Shopify, Ramp, Instacart, Gusto, and Samsara use it. Agents work across 400+ connectors, drive a real browser, answer in Slack, Teams, email, and iMessage, and review their own runs to get cheaper over time. The original canvas workflows are now labeled legacy on its pricing page.
General Input shares the goal of letting the people who understand the work automate it. The difference is the shape of what gets built. Each General Input specialist owns one recurring process and turns it into skills, workflows, and apps your team can review. Fixed steps run as code that behaves the same on every run, judgment steps use a model, and the price is per seat rather than a fee on top of every token.
What Gumloop does well
- Agents anyone can build. Non-technical teams set up agents without learning a canvas, and an agent one person builds can be shared with the whole company.
- A strong control plane. Spend by agent, model, and team, spend caps with approvals, agent health, and evaluations that grade runs against your rules.
- Browser agents done carefully. Saved logins and a password vault, handoffs when a site needs a person, and recorded sessions with masked passwords.
- Enterprise options. SOC 2 Type II, RBAC, audit logs, SCIM/SAML, and VPC deployment in your own cloud.
How they compare
| General Input | Gumloop | |
|---|---|---|
| Unit of workGumloop: agents, shared across the company. General Input: a specialist per recurring process, with its own skills, workflows, and apps. | Specialists | Agents |
| Natural language building | ||
| Deterministic code workflowsGeneral Input runs fixed-step jobs as code with no model in the loop. Gumloop's canvas workflows are now labeled legacy. | ||
| App integrationsGumloop: 400+ connectors plus custom MCP. General Input: 850+ apps with 80,000+ documented operations, plus custom API connections. | 850+ | 400+ |
| Browser use with saved logins | ||
| Chat in Slack and Microsoft Teams | ||
| Human-in-the-loop approvals | ||
| Pricing modelGumloop Pro: $37/month for 20,000 credits, billed at list cost plus an 8% fee (16% with your own model keys). General Input Pro: $30 per seat per month with 4,800 credits. | Per seat | Usage + fee |
| Per-run cost breakdown | ||
| Least-privilege credential scopingAgent Firewall limits each credential to specific operations, enforced outside the model. Gumloop's connector policies are per agent on Pro and org-wide on Enterprise. | ||
| Self-hosted / on-prem deploymentGumloop Enterprise can run in your own AWS, Azure, or Google Cloud account. General Input Enterprise deploys on-prem. |
Where they differ
Agents or processes?
Gumloop centers on agents: you build one for a job, share it, and people put it to work in their own chats. General Input centers on the process. A specialist for collections, for example, holds the skill that describes how your team chases invoices, the workflow that runs every morning, and the app where someone marks what was handled. Anyone on the team can open each piece, see what changed, and pick it up when the person who built it moves on.
What does a run cost?
Gumloop bills the list cost of every token, compute second, and paid tool call, plus an 8% orchestration fee on that total. Bring your own model keys and the fee rises to 16%. Pro is $37 a month for 20,000 credits at $0.005 each, the allowance resets monthly, and when it runs out agents stop until pay-as-you-go is switched on. General Input Pro is $30 per seat per month with 4,800 credits. Code workflows cost less than one credit per run, and overages bill at the plan's rate so nothing stops mid-month.
How much of the work is exact?
An agent decides its steps each time it runs, which is right for research, drafting, and triage. A lot of business work is not that: move the paid invoices, post the numbers, update the CRM. General Input runs those steps as code that does the same thing every time, and saves agent reasoning for the parts that need judgment. You choose per workflow, and you can read either kind before it runs.
When to use Gumloop
Choose Gumloop when you want a company-wide agent platform with detailed spend analytics, evaluations, and browser agents, and your security team prefers running it inside your own cloud account. It suits AI-forward companies rolling agents out to many teams at once.
When to use General Input
- The work is a recurring process. Specialists own reporting, collections, research, and onboarding end to end.
- Some steps must be exact. Code workflows for fixed steps, agents only where judgment is needed.
- You want predictable pricing. A flat price per seat with credits included, and no fee on top of model costs.
- Credentials need tight limits. Agent Firewall scopes each connection to specific operations.
- You need it on-prem. Enterprise deployments can run on your own servers.
FAQs
Is General Input a Gumloop alternative?
Yes. Both let non-technical teams automate work with AI. General Input organizes that work into specialists that own recurring processes, runs fixed steps as deterministic code, and prices per seat instead of adding a fee to every task.
Does General Input charge a fee on top of model costs?
No. Plans include a monthly credit allowance per seat, every run shows its credit cost, and pay-as-you-go credits are $0.0069 each beyond that. You can also connect AI subscriptions you already pay for.
Can I still build visual workflows?
General Input does not have a drag-and-drop canvas. You describe the workflow, review what the platform builds, and edit it in plain language or code.