Frequently asked questions

Four routes, and you pick. (1) Your own AI client over MCP — point Claude Code, Codex CLI or Claude Desktop at your org and it picks up around sixty org-grounded tools, with the model running on the AI subscription you already pay for. (2) Your own provider key — Anthropic (Claude), OpenAI, OpenRouter, Ollama Cloud, Cerebras or any OpenAI-compatible endpoint, billed to you directly by that provider. Neither of those costs anything on top of your plan, and nothing is metered here. (3) AI credits from us, on Scale (50,000,000 tokens per user per month) or Max (200,000,000), if you would rather not manage a provider account at all. (4) None — a large part of the platform never calls a model, so the Metadata Explorer, deployments, the Permissions Workbench, the Data Loader, File Backup and the deterministic modules of Security Review all work with no AI whatsoever.

On Pro, AI features run on an API key you supply, from Anthropic (Claude), OpenAI, OpenRouter, Ollama Cloud, Cerebras, or any OpenAI-compatible endpoint. You add it once in Settings; it is encrypted at rest and used only for your own requests. You pay that provider directly for what you use. Scale and Max include platform-funded credits instead, if you would rather not manage a key.

If a request fails midway, times out, or produces a result you cannot use, the tokens it consumed are still billed by your provider and cannot be refunded. Create a dedicated API key with a monthly spending limit set at your provider, and keep an eye on that provider’s usage console.

No — admins are one of four groups it is built for. Consultants connect as many client orgs as they like, each isolated with its own metadata and context, and document an unfamiliar one in an afternoon. Developers use the Development Studio pair programmer, org-to-org diffs and selective deployments, or drive the same tools from Claude Code and Codex over MCP. Data analysts describe a report or dashboard in plain English, write SOQL in natural language, and edit live records in a spreadsheet view. Architects, RevOps and release managers use it too. One connection, one per-user price, and the same tools underneath for all of them.

No packages and no code. You create a Connected App in Setup (a five-minute, clicks-only task), paste its Consumer Key and Secret into orgadmin.ai, and authorise with OAuth. Your Salesforce password is never entered into orgadmin.ai.

Any edition with API access — Enterprise, Unlimited, Performance, and Developer Editions, plus all sandbox types. Production and sandbox orgs can be connected side by side.

Yes. A workspace can hold as many org connections as you like (production, sandboxes, and client orgs) each isolated with its own metadata, context, and AI learnings.

Turn your workspace into a company workspace and invite your team by email or straight from a Salesforce org. There are owner, admin, member and read-only roles, plus a separate “Deploy & write” toggle per member. Each person connects orgs with their own Salesforce login. Owners and admins see team activity and every deployment and Studio project, and removing someone disconnects every org login they made. Org context and AI-learned knowledge are shared across the workspace; favourites and saved queries stay personal.

Pro is A$14.99 per user per month and includes every feature in the platform, with the AI running on an AI provider key you supply. Scale (A$32.95) and Max (A$69.95) are the same product with platform-funded AI credits included, so there is no provider account to manage. Every plan starts with a free 14-day trial, and there is no lock-in contract.

No. Every plan is paid, and every plan includes the whole product — the Metadata Explorer, deployments, the Permissions Workbench, the Data Loader, data seeding, File Backup, Security Review and every AI feature. The only difference between the tiers is where your AI tokens come from. New accounts get a 14-day trial with no credit card so you can try all of it first.

Tokens measure how much text the AI reads and writes. On Pro you bring your own provider key, so your provider bills you directly and nothing is metered here. Scale includes 50,000,000 platform-funded tokens per user per month and Max includes 200,000,000. Those are Flash-equivalent tokens: each request draws what it actually cost, expressed in tokens of our standard model, so work the AI provider serves from its cache or runs in the background draws less than its raw size, and features running on a heavier model draw more. Usage is visible on your usage page at all times, and anything you run on your own key never draws it down.

Per member. A company workspace has one subscription on one plan, and its seat count follows the number of members: add someone and a seat is added, remove someone and it comes off, prorated either way. Pending invitations are not charged. The owner, and any admin given the billing toggle, manage the plan and invoices.

Not to start. Sign up, connect an org, and use every feature for 14 days with no card. A card is only needed when the trial ends and you pick a plan.

No. There is no lock-in contract, no minimum term and no platform minimum — you pay for a month at a time and stop whenever you like. That is deliberate: a customer who can leave in thirty days is one we have to keep earning, which is what keeps the features shipping and the price defensible.

Yes. Subscriptions are month to month, and you cancel yourself from your billing settings — no phone call and no cancellation fee. Your plan keeps working to the end of the period you have already paid for, and nothing renews after that. Disconnecting an org also revokes its token at Salesforce and purges the stored metadata and cached dashboard results.

Yes, when you ask it to — it can deploy metadata, edit permissions, load and edit records, and seed sandboxes. What it never does is change something while you are merely looking around: browsing, chat, search, and analytics run through a client where create, update, delete, and deploy methods do not exist. Every write feature needs the “Deploy & write” permission (owners and admins always have it; members only when an admin grants it), is disabled unless your deployment enables it, and confirmed by you before it runs — with production deploys requiring an explicit per-run confirmation.

OAuth refresh tokens and Connected App secrets are encrypted at rest with AES-256-GCM, never logged, and never sent to the browser. Every Salesforce API call is audit-logged — without tokens.

No. Prompts are sent to the AI provider your workspace admin chooses, under that provider’s API terms, and orgadmin.ai never trains models on your data.

Only briefly, so a dashboard does not re-query your org every time it is opened. The most recent run is kept for the dashboard’s refresh window (15 minutes by default, 24 hours at most), encrypted with AES-256-GCM under its own key, tied to the Salesforce login that read it, and left out of backups. Set a dashboard to Always live and nothing is kept.

orgadmin.ai revokes the token at Salesforce, deletes the stored secrets, and purges all cached metadata and cached dashboard results for that connection.

No. orgadmin.ai is an independent product, built in Australia, and is not affiliated with, endorsed by, or sponsored by Salesforce, Inc.

Agentforce is Salesforce’s platform for building customer-facing AI agents, licensed from US$125 per user per month at list. orgadmin.ai is an AI workspace for the people who work in the org (admins, consultants, developers, and analysts) starting at A$14.99 per user per month. See the full comparison on our Agentforce page.

Your choice: Anthropic (Claude), OpenAI, OpenRouter, Ollama Cloud, Cerebras, or any OpenAI-compatible endpoint. On Pro you bring your own key, and models are configured per feature by your workspace.