Automated user provisioning without SCIM is possible: invite people through each tool's member API, let your company domain do the joining where a tool supports it, script those steps around one trigger in your directory, or use a browser agent that works each tool's admin page for you. None of these is as clean as SCIM. All of them work on the plans small companies actually pay for.
User provisioning covers what provisioning is and why doing it by role matters. This post is about the "automated" part, for the common case where SCIM is not on the table, and about how to automate without copying mistakes to every new hire.
Why SCIM is not an option for most small teams
SCIM is the standard API an identity provider uses to create, update and remove accounts in your tools. It needs two things to line up, and for a small team they rarely do.
First, you need an identity provider that does provisioning, which is usually its more expensive tier. Second, every tool you want covered has to offer SCIM on the plan you are on. Most keep it on Business or Enterprise plans, and many require SAML single sign-on to be set up first. The table in What is SCIM? shows how that plays out tool by tool; SCIM provisioning adds the setup and costs.
Slack is a clean example of the gap. Its developer docs say the SCIM API is only available to workspaces on Business+ or Enterprise plans, as we read it on 27 October 2026. A company on Slack Pro has no SCIM to call, whatever identity provider it buys.
The result is that a 20-person company on standard plans can usually get SCIM for a couple of its tools at most. Upgrading the rest to reach it is the SSO tax under another name. Everything else needs another route.
Member APIs
Most tools publish an API for managing members that is separate from SCIM, and some offer it on every plan.
GitHub, for example, documents "Create an organization invitation" (POST /orgs/{org}/invitations) in its REST API, with no plan restriction stated. The caller must be an organization owner. The call takes a role (member, admin or billing manager, with member as the default) and a list of team IDs, so one request can invite an engineer straight into the right teams. GitHub's docs also set a ceiling: organizations older than a month or on a paid plan can send 500 invitations in 24 hours, others 50. That is plenty for hiring, and a reminder that every API has limits you only find in the small print.
Other vendors keep their member-management APIs for the top plan. Slack's admin.users.invite method, which invites a user to a workspace, says it is only available on an Enterprise plan, as we read it on 27 October 2026. So in Slack, the API route and the SCIM route are both closed below the expensive tiers, and you are left with domain joining or the admin page.
When you look at a tool's member API, ask four questions. First, can it invite, or only read? Some APIs list members on every plan but only let you change them on a higher one. Second, can it set the role? An invite at the default role is half the job. You want "editor in this project", not "member somewhere". Third, does an invite cost a seat before it is accepted? Pending invitations sometimes count, so know before you send ten. And fourth, where does the token live, and who owns it? An admin token is the keys to that tool. It should belong to the company, not to whoever wrote the script, and it should not sit in a file on someone's laptop.
Invite flows and domain capture
Several tools let anyone with your company email domain join on their own. It is automation of a sort: nobody sends an invite.
Slack lets workspace owners and admins approve email domains so that anyone with an approved address can join, on the Free, Pro and Business+ plans, per Slack's help center as we read it on 27 October 2026. Notion has "allowed email domains" in workspace settings. Per Notion's help page, people with a matching email see the option to join, become members rather than guests, and are billed. The page does not say which plans have the setting.
Domain capture is convenient for tools everyone uses. It has three weaknesses. Everyone gets the default role, whatever they need. You have no record of who joined when unless you go and look. And it does nothing at the other end: people who joined on their own are not removed on their own. That makes it a fine choice for chat, and a poor one for anything with customer data, money or production access.
Just-in-time provisioning is the SSO version of the same idea: the account is created on the first login through your identity provider. It has the same weakness at the leaving end, and it needs SSO on the tool's plan. Just-in-time provisioning covers when that is good enough.
Scripts and triggers
A script ties the pieces together. When a new person appears in Google Workspace, it calls the GitHub invitation API, adds them to the right Google Groups, and posts a checklist for the tools it cannot reach. The best part of a script is usually the trigger, not the invites: one event (a new account in the directory) that starts provisioning without anyone remembering.
Scripts age badly. APIs change, tokens expire, and the person who wrote it moves on. If you write one, keep it short, log every action, and make failures loud. A script that silently skips one tool is worse than no script, because everyone assumes it worked.
Workflow tools such as Zapier are the no-code cousin. They are good at the trigger (a new row in the hiring sheet, a new user in the directory) and limited by each app's integration, which often has no "invite to workspace" action at all. Check the action list for each tool before you design around it.
Browser agents
For the tools with no member API on your plan, and no domain join, the last option is software that uses the admin page as a person would: open members, click Invite, type the email, pick the role, send.
This is how SmashSSO handles tools without a token: an agent works in your own Chrome, through the SmashSSO extension, in tabs where you are already signed in as an admin. It covers tools that have no usable API on your plan, because it uses the same screens you would.
It comes with limits. Chrome has to be open for those runs. They are slower than an API call. They can fail when a tool redesigns its screens, and a failed run says so and shows its last screenshot. Some tools' terms restrict automated use, so check the ones that matter to you.
A worked example: one engineer, eight tools
Take a 20-person company hiring an engineer. The bundle says Google Workspace, Slack, Zoom, the password manager, GitHub (member of the web team), Linear, Vercel and Sentry.
Google Workspace is where the person is created, by hand or from the HR system; that is the trigger. Slack is on Pro, so there is no SCIM and no invite API, but the company domain is approved, so the new hire joins with their work address on day one. GitHub has a member API on the plan, so the invite goes through it with the role and team set. For Linear, Vercel and Sentry, the company checks each one's API reference for a member endpoint on its current plan and uses it where it exists. Anything left is done in the admin page, by a person with a checklist or by an agent.
The result is not one mechanism but four, chosen per tool. That is normal. What makes it automated is that one event starts all of them, and one record says what each produced.
How to handle paid seats
Automatic invites to free tools are low risk. Automatic invites that add a paid seat are a spending decision, and should stop and ask, at least until you trust the bundles. Check whether each tool bills pending invitations, because an invite nobody accepts can still cost you. And when someone leaves or changes role, take the seat back as well as the access; a deactivated account on a seat you still pay for is only half done. SaaS license management covers the money side.
What to automate first
Automate in the order that saves the most clicks with the least risk.
Write the bundles first. Decide what each role gets, tool by tool, with the role in each. Automation without a definition only makes inconsistency faster.
Then the everyone bundle. Chat, video calls, the password manager, the HR portal. Every new hire needs them, so they repay the effort first.
Then your biggest role. If half your hires are engineers, the engineer bundle comes next.
Leave the rare tools manual. The tool two people use can stay a checklist line, as long as it is on the list.
Keep a human on paid seats, as above, until the bundles have been right for a few hires in a row.
A five-person company can stop after the first two steps. A fifty-person company hiring every month will feel the difference from each one. How to automate onboarding turns this into a first-day routine.
Is automated provisioning safe?
Mostly, if the bundles are right. The risks are giving too much (an admin role in a bundle by mistake, copied to everyone), inviting the wrong address (a typo that sends an invite to a stranger), and losing track of what was given. Match on the email in your directory, review bundles when roles change, and keep a log. Provisioning also creates the list you will need later for removal: provisioning and deprovisioning explains why that record matters, and automating offboarding covers the other end with the same tools.
Where SmashSSO fits
SmashSSO does role-based provisioning without SCIM. It reads your people from Google Workspace with read-only directory access. You define access bundles such as "Engineer", and it invites a person to every tool in the bundle with the right role: through a tool's member API where you have added a token, otherwise through the agent in your Chrome. Any step that would add a paid seat pauses for your approval, and every change is recorded in an audit log. Role changes and removals run the same way.
It is not SSO, it needs Google Workspace as your directory, and it has no SOC 2 or ISO 27001 report. It costs a flat fee from $199 a month, billed annually, however many people and tools you have (pricing). How it works walks through a bundle, and the tools page lists what it covers.