Proudly built and operated in Australia
Ask it anything about your Salesforce org and it goes and looks. Then it can act: dashboards, permission edits, data loads, Apex and Flows. For admins, consultants, developers and analysts — and you confirm every change before it lands.
OAuth in about five minutes. No managed package to install.
This chat: 41,208 tokens
Six things are wired to Opportunity. Let me read the flow and check what else mentions the stage by name.
Gathered context · 2 stepsRead flowOPPORTUNITY - Start Credit Assessment · v7 · 23 elementsSearched metadatafound 4 matching "Closed Won"Every answer is read out of your live metadata, in your org, at the moment you ask.
The assistant runs about two dozen investigation tools before it replies, so it is describing the org you actually have rather than the one Salesforce documents. Nothing it finds changes anything. Writing is a separate system with its own gates, and it asks first.
One connection, one price, and the same twenty-odd investigation tools underneath. What changes is which part of the job you brought with you.
You run the org and field every “why did this happen?” question in it.
You work across a dozen client orgs and have to be useful in an unfamiliar one by Thursday.
You are writing Apex and LWC against an org somebody else built.
You need the number before the 2pm call, not a ticket in the admin queue.
Also here: Architects, RevOps and ops leads, release managers, and the accidental admin who inherited the org along with the rest of the job.
OAuth against production or a sandbox. Nothing is installed in the org and no password is shared.
Two dozen org-aware tools run against your metadata, your automation and your live data before you get an answer.
Deploy, edit permissions, load records. Each one is reviewed and confirmed by you first.
One workspace for your admins, developers and contractors, on one subscription billed per member. Everyone connects with their own Salesforce login, and the people in charge can finally see who shipped what, to which org, and when.
Owner, admin, member and read-only roles, plus a separate “Deploy & write” toggle, so a developer can ship without also managing people or billing.
Every deployment and validation in one team log: who ran it, which org, how many components and how it went. Plus every Studio project and its versions.
Two people changing the same component, in any sandbox, are flagged as they work. A deploy that would undo someone’s work waits until you confirm you have checked with them.
Remove a member and every org connection they made is disconnected, with the token revoked at Salesforce. Their history stays with the team.
The AI in orgadmin.ai is not a licence we resell you. There are four ways to power it, and two of them cost you nothing beyond the plan itself.
Every plan, including the trial
Point Claude Code, Codex CLI, Claude Desktop or any MCP client at your org and it picks up around sixty org-grounded tools: read metadata, run SELECT-only SOQL, audit security, build reports and drive the Development Studio.
The model runs in your client, on the AI subscription you already pay for. Nothing is metered here and no allowance is drawn down. Each token you mint carries only the scopes you grant it, starting read-only.
Every plan. It is how Pro works.
Add a key from Anthropic (Claude), OpenAI, OpenRouter, Ollama Cloud, Cerebras or any OpenAI-compatible endpoint. Every AI surface in the app then runs on it, and you can point each feature at a different model.
Your provider bills you directly at their rates — we never meter that usage or mark it up. Failed runs still consume tokens, so set a monthly spending limit on the key you use here.
Scale and Max
No provider account, no key, nothing to set up. Scale includes 50M tokens per user per month and Max includes 200M, funded by us and working the moment you sign in.
Usage is on your usage page at all times. Connect your own key as well if you like — own-key usage never draws down your included credits.
Every plan
A large part of the platform never calls a model. The Metadata Explorer, org comparison and deployments, the Permissions Workbench, the Data Loader, File Backup, the Formula Tester and the debug-log viewer are ordinary deterministic tools.
Security Review runs its fourteen deterministic modules with no AI whatsoever; only the executive summary and the judgment findings need a model. You can get real work done here without a key, an MCP client or a single credit.
Every plan carries the whole platform. Pro runs the AI on your own key or your own MCP client; Scale and Max include credits instead. One flat per-user price, published here, with no platform minimum and no lock-in contract — it is month to month, so we have to keep earning it.
| What the alternatives charge | Per user / month |
|---|---|
| orgadmin.ai Pro | A$14.99 |
| Agentforce user add-on | US$125 |
| Gearset Deployments, Starter | US$215 |
| Agentforce 1, per-user edition | US$550 |
List prices read from each vendor’s own pricing page or marketplace listing on 19 August 2026. USD converted at A$1 = US$0.7049 (Reserve Bank of Australia reference rate, 19 August 2026); AUD figures are indicative, rounded, and exclude GST, implementation, support and negotiated discounts. Packaging differs between vendors and most figures above are entry tiers that exclude modules needed to match this platform’s scope — obtain a current quote before making a purchasing decision. All product names and trademarks are the property of their respective owners.
Everyone who works on orgadmin.ai has run a Salesforce org. That is why it starts from the hundred small investigations that fill the day of anyone working in an org: what fires when this Opportunity closes, where this field is referenced, why this validation rule exists and who to ask about it. Each of those used to be twenty minutes in Setup.
AI is very good at exactly that job. The catch was the licensing: per-user add-ons at enterprise prices, consumption credits to budget for, and free developer tooling moving to paid tiers in 2026. Trying AI on your own org became something you needed procurement approval for.
So we built one that doesn’t. AI for every admin, consultant, developer and analyst — at a price that never needs a business case.
Dashboards now work like Salesforce’s own: they show when their numbers are from, reuse a recent run instead of re-querying your org on every open, and refresh live whenever you ask.
Embedded dashboards are easier to roll out to everyone, safer across sandboxes, and clearer when something is set up wrong.
The pricing page currency switcher now also covers British pounds, Swiss francs, Japanese yen, and Singapore and Hong Kong dollars. Billing stays in Australian dollars.
Search for Claude and Salesforce and you'll find four different things: a Salesforce plugin inside Claude, Salesforce's own MCP servers, Claude running inside Agentforce and Slack, and MCP servers built for admins. Here is what each one does, who it's for, and how to connect it safely.
Setup deactivates one user at a time, and the 'You cannot deactivate this user' error stops you halfway. Here is a safe order for offboarding many users at once: freeze, clear the blockers, deactivate in bulk, then reassign what they owned.
Getting every file out of Salesforce is harder than getting the records out. Here are the four ways to do it, what each one actually exports, the limits that catch people out, and how to keep the link between each file and its record.
Every AI feature from A$14.99 per user per month, on your own key, your own MCP client, or credits from us.