FAQs
What happens when I install Swantide?
Swantide is a managed package with a connected app. After the package is installed and authorized in an org, Swantide pulls down metadata via the Metadata & Tooling API. We also use these APIs for metadata deployments via AI Admin and Workflows.
How do I get an org documented? What is actually happening when you document an org?
Swantide uses the Salesforce Metadata API and Tooling API to ingest and generate documentation for every metadata component in Salesforce.
If you need an org to be connected to Swantide, reach out to our team. We will create a Swantide instance, which will include an onboarding form with a link to install the Swantide package and authorize the connection.
How many users can I add to Swantide?
Swantide is priced on a per-org basis. This means that you can add as many users as you want to have access to Swantide. If you want to add users, send their email address to Swantide and we will create them as users. We are working to add user permissioning in app for customers to manager
What are the prerequisites for using AI Assistant & AI Admin?
The Salesforce org needs to be documented by Swantide. Note that the process of documenting all of the metadata in an org takes 24 - 48 hours, so there is some lead time required for this.
How do I add users / manage permissions? How can I access different orgs?
If you have multiple orgs connected to Swantide, you can change the org that you’re in by going to Profile. This will show all of your available orgs.
Today, user and org management is handled by the Swantide team. If you need to add new users, deactivate users, or add new orgs, send the updates to Swantide. Soon we will have user management in app to be managed by customers.
Can Swantide support things like CPQ that have managed package components and/or use record-level data as part of business logic?
Yes! Swantide can see all of the metadata that is not hidden by a managed package and not record-level data. We can see metadata that has been built to reference package components as well. We have several customers who use Swantide in orgs with CPQ to understand the custom logic, write test cases, write enablement documentation, etc.
On our product roadmap is the ability to opt into having certain record-level data documented and in scope for Swantide.
What visibility do different users have?
Documentation in the data dictionary is shared across all users in the org
AI Admin tasks are visible to all users in the org, but can only be deployed by the user who created them
AI Assistant chats are stored at the user level
What AI models does Swantide use?
Swantide leverages enterprise grade models from OpenAI, Anthropic, and Gemini. The Swantide product is built in a way that allows us to swap out models as needed.
Are there any technical limits to the Swantide AI Admin and AI Assistant?
Two important rules of thumb when prompting the AI Assistant:
The AI Assistant can handle 10 - 15 metadata components per thread. When you ask a question, Swantide searches for the most relevant metadata to retain as context to answer the question. This helps reduce time to response and prevent diluted answers.
If you need to answer a question that references more than 10 - 15 metadata components, use the References table in the data dictionary. This table is an exhaustive list of metadata references, and you can filter from there
For each prompt to the Swantide AI Assistant, expect about a 1 page answer
The Swantide AI Admin can handle 40+ metadata components as part of a single task. If you need to update that many components as part of a change you’re making, it may be easier to break down the task into smaller chunks, depending on the complexity and number of revisions the task requires.
Neither products touch anything that is record-level data; the scope is exclusively metadata.
Swantide does not support file uploads - everything must be included in the body of the prompt.
How can I filter in the data dictionary?
You can filter by metadata type (e.g. CustomObject, CustomField, Flow, Validation Rule). We leverage AI to compile this list of references by searching through the metadata JSON of every component in the org.
If you want to filter to Flows that reference a specific object or field, find that field in the data dictionary, navigate to references, go to ‘Components that reference this component’, and filter to Flows.
An important consideration is that these tables show all references
If you need a more granular list (e.g. all flows that run when an opportunity is created), today we suggest finding the list of all references in the Data Dictionary and asking more targeted questions
What do I do if there is a component that isn’t documented?
Swantide can see what you see in the org - Salesforce managed packages that include Apex Code are not visible in Salesforce or to Swantide. Other managed package components (e.g. objects, fields, validation rules, etc.) that are not hidden are visible to Swantide. The apex classes will appear in the data dictionary, but there will be no documentation generated.
What Salesforce products does Swantide support?
Generally, everything on the core Salesforce platform is accessible through Swantide and uses the metadata API. Products that are not on the core Salesforce platform (legacy Pardot, Tableau, etc.) are out of scope for Swantide.
Can I export the documentation?
Since documentation is updated nightly, exporting can lead to having inaccurate documentation. Best practice is to refer to documentation in the Swantide App. A copy/paste function can also be used in AI Assistant for specific pieces of documentation. This is also useful
Does Swantide have access to old versions of metadata components?
Swantide does maintain a change log, accessible through the data dictionary, that shows the date and time of the last change.
The AI Assistant has access to the current version of documentation and the version immediately prior. We don’t store older documentation to assist with performance.
You can ask Swantide what changed in the prior version, and it will summarize for you
How often is the documentation updated? How does that work?
After initial documentation (24-72 hours), orgs will be updated every night to reflect any new changes
How does Swantide retain context in an AI Assistant thread?
Swantide is able to retain context in an existing thread. This means that you shouldn’t need to continue referencing the names of the same metadata components when asking follow up questions.
Because we retain the context, you should always start a new thread if you’re asking a different question. This helps prevent retaining irrelevant context that could dilute your answer.
What happens if I want to manually sync an update before the nightly sync?
Using the Data Dictionary, select the metadata component you would like to update. Once selected, click on the “Sync Component” button on the top right of your screen.
This works for syncing existing metadata components. We are releasing a feature very soon to be able to sync new metadata components instantly.
How do I get support?
Reach out to the Swantide team on Slack channel or email (support@swantide.com)
What are common mistakes people make?
Asking AI Assistant too broad of questions or asking too much in a single prompt.
Ex: Can you show all of the active flows in my org?
Ex: Can you compare and contract all of the profiles in my org?
Not asking follow up questions to AI Assistant to continue to refine and get better answers
Tasking AI Admin with too much metadata (break up into smaller tasks)