Plauti offers several AI‑assisted features across its data management apps, and more are on the way. These features help you design better deduplication logic, review and merge duplicates more efficiently, and apply AI at scale to analyze or generate content based on your Salesforce data and attachments.

Quick reference – Which AI feature should I use?

Use the table below to quickly choose an AI feature based on your goal.

Data handling by the different AI features

Plauti’s AI features fall into two categories:

Plauti‑managed AI features (DPA‑based)

• Features: AI Merge Recommendation, AI Match Recommendation, AI Scenario Builder in Plauti Deduplicate.
• Require a signed Data Processing Agreement (DPA) for your Org or Environment.
• AI processing is limited and transient: only relevant data is sent over SSL to Plauti’s AI sub‑processor, used once, and then removed.
• Data is not stored by Plauti or its sub‑processors and is not used for model training.
• The AI only processes data requested by a user with the appropriate permissions; it cannot independently access your systems.
• AI suggestions are always visible and require your confirmation before any changes are applied, unless explicitly stated otherwise.

User‑managed AI features (your own LLM / provider)

• Features: Ask AI, Ask ChatGPT, Ask ChatGPT Vision in Plauti Manipulate.
• You configure and manage your own AI provider connection (OpenAI, Azure OpenAI, Google Gemini, AWS Bedrock).
• The prompts, field values, selected records, and any attached files you include in an action configuration are sent directly from Salesforce to the AI model you selected.
• The AI model can only see:
  • The main and follow‑up prompts you define
  • The fields and content you explicitly reference via Field Variables
  • The selected records (and optionally related records) processed by the action
  • Any files included via the action configuration
• The AI model cannot query other records on its own or bypass your configured field selection.

For more details on data handling by Plauti-managed and User-managed AI features, see AI Features and Data Processing.

Plauti‑managed AI features

These features require a signed DPA before they can be enabled. 

AI Merge Recommendation

AI Merge Recommendation provides AI‑generated, contextual field merge proposals in the Manual Merge modal. It analyzes all selected records holistically, and suggests which field values to keep on the master record, highlighting them for review and providing an explanation per field. No changes are made until you confirm them.

Use AI Merge Recommendation when:

• You already use Manual Merge but find the standard merge rules (e.g. “most recent”, “shortest value”) too limited.
• You spend significant time comparing and correcting values (typos, inconsistent formatting, etc.) across multiple records.
• You want AI to propose better default values but keep humans in control of the final merge.

Read more in AI Merge Recommendation.

AI Match Recommendation

AI Match Recommendation analyzes Deduplicate Job results, and labels each duplicate pair as Duplicate, Not Duplicate, or Uncertain, with an explanation. It considers the full record context, instead of just the fields and logic defined in the job’s scenario. Use the labels to decide whether to merge records or not; manually or with an auto merge job based on the AI advice.

• Catch duplicates that strict rules may miss
• Have uncertain pairs flagged even if the scenario score is high
• Reduce manual review effort by focusing attention where it is most needed

AI Match Recommendation can be used as part of a new Deduplicate Job, and for existing job results. You can also combine it with Auto Merge: for example, use AI status “Duplicate” plus a match score threshold (e.g. ≥ 97–98%) to auto merge with higher confidence.

Use AI Match Recommendation when:

• Your scenarios miss fuzzy duplicates, or produce many potential duplicates that require time‑consuming manual review.
• You want AI to help you decide which record pairs/groups are truly duplicates, particularly in large jobs.
• You want an extra safety signal before auto merging below 100% matching scores.

Read more in AI Match Recommendation.

AI Scenario Builder (Sandbox only for now)

AI Scenario Builder is an AI‑driven conversational assistant that helps you create or improve deduplication scenarios. It reads field metadata for the selected object (which fields are in use, types, etc., so not field content), asks clarifying questions, and proposes a scenario configuration (fields, weights, threshold, etc.) that you can review and adjust.

You can ask the assistant to help create a new scenario from scratch, as well as review issues with existing scenarios (e.g. too many false positives, missed duplicates) and suggest improvements.

After the assistant proposes a scenario, you can confirm its settings and save it in your Scenario settings.

Use AI Scenario Builder when:

• You’re new to Plauti Deduplicate and unsure how to design effective scenarios.
• You want help tuning scenarios that either miss duplicates or create too many false positives.
• You prefer to keep final control over scenario settings but would like AI guidance based on your object’s structure and your business requirements.

Read more in AI Scenario Builder.

User‑managed AI-features

These AI actions are part of Plauti Manipulate. Instead of Plauti managing the AI endpoint, you connect directly to your chosen provider. This gives you maximum flexibility and control over model choice, pricing, and data policies.

Ask AI action

The Ask AI action lets you send prompts plus Salesforce record data to an AI model, and store the response back on those records. It supports multiple providers: Azure OpenAI, Google Gemini, AWS Bedrock, and OpenAI (ChatGPT).

You can:

• Use field variables to inject values from the record and related records into the prompt.
• Run the Ask AI action from list views or as a Manipulate Job.
• Save the AI’s responses into text fields or picklists for further processing.

Some use case examples:

• Classify Contacts into departments based on their job title.
• Summarize related Cases per Account into an “Account Case Summary” field.
• Generate sales pitches, renewal summaries, or tailored descriptions based on record data.
• Analyze related Opportunities per Account and create financial summaries for executives.

Use Ask AI when:

• You want to decide for yourself which AI provider should be used.
• You want to integrate AI into batch processes, jobs, Flows, or Apex while retaining full control over the provider and prompts.
• You are starting new AI use cases in Plauti Manipulate: Ask AI is the recommended baseline going forward.

Read more in Ask AI action.

Ask ChatGPT action

The Ask ChatGPT action is a provider‑specific version of Ask AI, focused on OpenAI’s ChatGPT models. Functionally it is identical to Ask AI: you define prompts, insert field variables, optionally attach related data, and write responses to a field.

Use Ask ChatGPT instead of Ask AI when:

• You already have working Ask ChatGPT macros or flows and wish to continue using them.

Read more in Ask ChatGPT action.

Ask ChatGPT Vision action

The Ask ChatGPT Vision action extends Ask ChatGPT with image analysis. It selects one attachment per record based on criteria you define, sends that image plus your prompt to a Vision‑capable OpenAI model (such as GPT‑4o or GPT‑4 Turbo with vision), and stores the textual answer on the record.

Some use cases examples:

• Verify whether a Case attachment (e.g. screenshot) matches the Case description and subject, and store “Yes/No” in a picklist field.
• Read restaurant receipts or invoices from images/PDFs and write summarized expense or invoice details to fields.
• Analyze damage report photos in insurance or service contexts and produce draft damage descriptions.

Use Ask ChatGPT Vision when:

• Image or PDF content is central to your process and you want AI to interpret it.
• You need to bridge between visual information attached to records and structured text for your workflows (routing, approvals, reporting, etc.).

Read more in Ask ChatGPT Vision action.

FAQ – Plauti AI features

Do I need to sign the DPA to use all AI features?
No. The DPA is required for Plauti‑managed AI features in Plauti Deduplicate: AI Merge Recommendation, AI Match Recommendation, and AI Scenario Builder.

For Ask AI, Ask ChatGPT, and Ask ChatGPT Vision, you manage your own AI provider connection; these do not require the Plauti AI DPA, but you must comply with your provider’s terms and privacy policies. 

Is my data used to train AI models?
For Plauti‑managed AI features (DPA‑regulated), data is:

• Deleted after processing
• Not stored by Plauti or its AI sub‑processors
• Not used for training AI models

For Ask AI / Ask ChatGPT / Ask ChatGPT Vision, training behavior is governed by the policies of the provider you configure (OpenAI, Azure OpenAI, Google, AWS, etc.). Please refer to their documentation for details.

Can the AI see all my Salesforce data?
No. The following applies to both categories of features:

• Only users with appropriate permissions can invoke the AI features.
• Only the data explicitly retrieved for the operation is sent:
  • For Plauti‑managed AI, this is restricted to what the feature needs (for example, selected records or field metadata).
  • For Ask AI / Ask ChatGPT / Ask ChatGPT Vision, only the fields, related records, and files you explicitly include in prompts and configuration are sent.
• The AI models cannot query other records or bypass your configured field selection.

Does AI ever apply changes automatically?
By default, Plauti’s AI features present suggestions that you can review:

• AI Merge Recommendation highlights suggested values in Manual Merge; you decide what to keep before merging.
• AI Match Recommendation labels pairs/groups to guide your decisions; it does not merge by itself.
• AI Scenario Builder proposes scenario settings; you must confirm them before saving, and you can always edit them afterwards.

You can build automation on top of AI outputs (for example, Auto Merge based on AI Match Recommendation status, or Flows using Ask AI results), but those automations are always under your control.

Should I use Ask AI or Ask ChatGPT for new projects?
For new configurations, Ask AI is recommended because:

• It supports multiple providers (Azure OpenAI, Google Gemini, AWS Bedrock, OpenAI).
• It gives you flexibility to change models or providers without redesigning your setup.

Use Ask ChatGPT primarily if you already have working Ask ChatGPT macros/workflows that you want to keep using.