Generate a Blueprint with AI

AI Data Access is configured per object type — Contacts, Companies, Accounts, Deals, and so on. There are three levels:

To use Generate with AI for an object type, AI Data Access for that object must be set to Non-PII or Data. If it is set to Meta, the AI can only see field names and types — it cannot analyze value distributions, identify patterns, or generate data-grounded Blueprint rows.

Non-PII is the recommended setting for most teams. It gives the AI sufficient access to produce high-quality suggestions and generated Blueprints while keeping personally identifiable fields out of the AI pipeline.

To configure AI Data Access:

1. Go to Settings > Fields and mark any fields containing personal data as PII. Fields marked as PII will be excluded from everything sent to AI service providers during the generation process. This does not affect Data Logic execution.
2. Go to Settings > AI and set the access level for the relevant object type to Non-PII or Data. Changing the setting requires an Owner or Admin user role. To review or update user roles, go to Settings > Users.

Learn more about managing how your data is used with Insycle's AI features.

The image above shows the Meta, Non-PII, and Data toggles for the Contacts object type all enabled (yellow/right), on the Settings > AI page.

Generate with AI is accessed through the Data Logic module. It can also be reached from the Blueprints page, which displays a prompt directing you to Data Logic.

From Data Logic:

1. Navigate to Data Management > Data Logic
2. Select your CRM account and object type from the dropdowns at the top of the module
3. Under 1. Pick Blueprint, select Generate with AI
4. Click the Generate button

From Blueprints:

1. Navigate to Operations > Blueprints
2. Click the Generate button
3. A popup appears explaining that Blueprint generation is done through Data Logic. Click Got it to close the popup
4. Navigate to Data Management > Data Logic and follow the From Data Logic steps above

The Generate Blueprint with AI popup opens and begins analyzing your CRM fields. The analysis runs through three steps, each shown with a status indicator. The analysis usually takes 2–3 minutes. A progress bar and time remaining estimate display while it runs. When all three steps are complete, the Use Case tab becomes active, and you can proceed to Step 2.

When field analysis is complete, the Use Case tab presents three paths for generating your Blueprint. Select the one that best fits your starting point:

• Suggested for your data — AI reviews your CRM field population and gaps and suggests Blueprint use cases based on what it finds. Use this path when you're not sure where to start and want AI to identify logic opportunities in your data
• Common use cases — Browse a library of predefined use case patterns such as territory assignment, ICP scoring, industry normalization, and renewal risk scoring. Use this path when you have a use case in mind and want a starting point tailored to a familiar pattern
• Build your own — Select your own input and output columns and choose a Blueprint category. Use this path when you know which fields you want to work with and want AI to generate the rows based on your own instructions

On the Use Case tab, click the heading of your chosen path to expand it, then proceed to the relevant section below.

Use Case Option: Suggested for Your Data

Suggested for your data analyzes your CRM field population and gaps, and presents Blueprint use case options based on what AI finds in your data.

1. On the Use Case tab of the Generate Blueprint with AI popup, click Suggested for your data to expand the section. The number of records and fields analyzed is shown at the top.
2. Review the suggested use cases. Each suggestion includes:
   • A title and two tags — a pink Domain tag and a yellow Category tag — that identify the business area and type of logic the Blueprint applies. For example, a job title classification suggestion would show Data Hygiene and Classify. See Understanding Domain and Category Tags above for a description of each domain and category.
   • Why We Suggest This — An explanation of the logic opportunity AI identified.
   • What We Found — A summary of the distinct field values and patterns observed in your data.
   • Input columns — The CRM fields that will be used for matching, drawn from fields that exist in your CRM.
   • Output columns — The fields that will be updated when a match is found. These may not yet exist in your CRM — see How Generate with AI Works for guidance.
3. Click a suggestion to select it — a Selected label appears at the bottom of the popup confirming your choice.
4. Click Next to proceed to the Guide AI tab, then continue with Step Three: Guide AI below.

Suggestions in this section can span any domain — Data Hygiene, Sales & Revenue, Marketing, Systems & Integrations, or Compliance — depending on what patterns AI identifies in your data. See Understanding Domain and Category Tags above for a description of each domain and the types of logic it covers.

Note: Multiple suggestions may appear. Review each one before selecting — the What We Found summary can help you assess whether the suggestion is relevant to your use case.

The Generate Blueprint with AI dialog walks you through a three-step process: Use Case, Guide AI, and Preview. In the Use Case step shown above, Insycle analyzed 98 Contact records across 636 fields and surfaced Blueprint suggestions based on field population and data gaps. Each suggestion includes a name, category tags, an explanation of why it was recommended, and the input and output columns the Blueprint will use. Here, Postal Code Geographic Derivation is selected — Insycle detected that the Postal Code field is 95% populated and can reliably derive City and Region values to fill gaps in those fields.

Use Case Option: Common Use Cases

Common use cases presents a library of predefined Blueprint patterns for frequently used logic scenarios. Patterns can span any domain — Data Hygiene, Sales & Revenue, Marketing, Systems & Integrations, or Compliance — and are tailored to the object type selected in Data Logic. See Understanding Domain and Category Tags above for a description of each domain and the types of logic it covers.

1. On the Use Case tab, click Common use cases to expand the section.
2. Review the available patterns. Each option includes a title, the pink Domain and yellow Category tags, and a brief description of the use case it addresses.
3. Click a pattern to select it — a Selected label appears at the bottom of the popup confirming your choice.
4. Click Next to proceed to the Guide AI tab, then continue with Step Three: Guide AI below.

The image above shows the Use Case step of the Generate Blueprint with AI dialog, which displays three sections for selecting a Blueprint starting point: Suggested for your data (collapsed here), Common use cases, and Build your own (also collapsed). The Common use cases section offers pre-built Blueprint patterns browsable by category, each displaying category tags, a name, and a brief description. Patterns shown here include Persona Mapping from Job Title, Lead Scoring by Geography and Engagement, Segmentation by Persona and Geography, Suppression by Company Name Pattern, Lifecycle Stage Enforcement, Consent Management by Geography, and Attribution Channel Normalization. Lead Scoring by Geography and Engagement is selected, as shown at the bottom of the dialog. Click Next to proceed to the Guide AI step.

Use Case Option: Build Your Own

Build your own lets you specify the input and output columns, select a Domain and Category, and use AI to generate rows based on your instructions. Use this path when you know which fields you want to work with and want AI to build the Blueprint structure for you.

1. On the Use Case tab, click Build your own to expand the section.

2. Select a Domain that describes the business area your logic belongs to. Selecting a Domain narrows the Category list to only the options relevant to that area:

   Domain	Categories Available
   Data Hygiene	Normalize, Number Band, Derive, Classify, Date Band, Other
   Sales & Revenue	Territory, Lead Routing, Scoring, Account Health, Pricing, Other
   Marketing	Persona, Segmentation, Attribution, Lifecycle, Other
   Systems & Integrations	Enrichment, Data Quality, Migration, System Sync, Product & SKU, Other
   Compliance	GDPR, Suppression, Consent, Other
   Other	All of the above categories

3. Select a Category that describes the specific type of logic you want to apply.
4. Select your Input columns — the CRM fields that will be used for matching.
5. Select your Output columns — the fields that will be updated when a match is found.
6. Click Next to proceed to the Guide AI tab, then continue with Step Three: Guide AI below.

Note: The Domain and Category you select are passed directly to the AI as context for generation — the more specific your selection, the more targeted the generated Blueprint will be.

The image above shows the Build your own option in the Use Case step of the Generate Blueprint with AI dialog. Rather than selecting a pre-built pattern, Build your own lets you define a custom Blueprint by selecting your own input and output fields. You choose a Domain and Category to classify the Blueprint — here set to Data Hygiene and Normalize — then select the fields Insycle will read from (input columns) and the fields it will write to (output columns). In this example, City, Country/Region, and State/Region are selected as input columns, with Phone Number and Mobile Phone Number as output columns.

The Guide AI tab appears after selecting a use case from any of the three paths. It shows the AI instructions that will be used to generate your Blueprint and gives you the opportunity to review and refine them before proceeding.

1. Under Choose a Variation, review the available variation or variations. Each variation includes:
   • A title describing the approach
   • Instructions for AI — the instructions that will be sent to AI to generate the Blueprint rows
   • Input columns and Output columns for that variation
2. Click a variation to select it.
3. Under Guide AI with your own words, review the pre-populated instructions. Edit the text to tailor the logic—for example, by adding specific values, adjusting the matching approach, or refining the output structure. If using the Build your own — Other category, this field is required and will be empty.
4. Click Preview to generate the Blueprint and proceed to the Preview tab.

  Note: The instructions in the Guide AI with your own words field are sent to AI along with your CRM field value distributions to generate rows grounded in your real data. More specific instructions generally produce more accurate and useful Blueprint rows. If you are using Build your own, this field is required — clicking Preview without entering instructions will display a validation error. For all other paths, the field is pre-populated with AI-generated instructions that you can edit or leave as-is before proceeding.

The image above shows the Guide AI step of the Generate Blueprint with AI dialog. This step displays the selected use case — here, Lead Scoring by Geography and Engagement — and asks you to choose a variation that defines how the Blueprint logic will be structured. Each variation shows the instructions that will be sent to the AI, along with the input and output columns it will use. In this example, Score contacts by state and number of associated companies is selected, using State/Region and # of Associated Companies as inputs to populate Priority and Score output fields. A second variation, Score contacts by job title seniority and geographic market, is also available. The Guide AI with your own words section at the bottom lets you edit the AI instructions directly before proceeding to the Preview step.

The Preview tab shows the Blueprint that AI has generated based on your use case, variation, and Guide AI instructions. Review it carefully before saving.

The Preview tab includes four sections:

• Blueprint Definition — the auto-generated name and description for the Blueprint. You can edit the name after saving from Operations > Blueprints
• Blueprint Structure — the input and output columns, showing which columns will be used for matching and which will be written to CRM fields
• Rows — the generated Blueprint data, paginated. Review the rows to confirm the logic looks correct, and the values are relevant to your use case
• Info panel — an informational note below the rows table that provides specific guidance about the generated Blueprint, including field value combinations that don't appear in your data and won't match any row, fields that are currently empty in your CRM, and recommendations for fallback values or additional configuration steps to consider before running. Review this note carefully before saving

If the Blueprint does not look right, click Back to return to the Guide AI tab, refine your instructions, and regenerate.

When you are satisfied with the Blueprint, click Save.

The Blueprint is saved to Operations > Blueprints and automatically loaded into Step 1: Pick Blueprint in Data Logic, ready for you to configure Input and Output Mapping.

The image above shows the Preview step of the Generate Blueprint with AI dialog, where Insycle displays the AI-generated Blueprint before you save it. The Blueprint Definition section shows the AI-drafted name and description. The Blueprint Structure section confirms the input and output columns — here, State/Region (Exact) and # of Associated Companies (Number Between) as inputs, with Priority and Score as outputs. The Rows section displays the generated logic as a searchable table, with each row representing a combination of input conditions mapped to output values. In this example, contacts in CA with 2–4 associated companies receive a High priority and score of 90, while contacts in NY or MA with 1–4 associated companies receive Medium priority and a score of 65. The Blueprint contains 41 rows in total. A notes panel at the bottom surfaces AI-generated observations and recommendations about the data. Click Save to save the Blueprint to Data Logic.

After saving, the generated Blueprint is automatically loaded into 1. Pick Blueprint in the Data Logic module. The Blueprint name is shown in the Pick Blueprint field, and a Preview (eye) icon appears beside it. Click the Preview icon to view the Blueprint table data without leaving Data Logic.

At this point, you have two options:

Option 1: Proceed Directly in Data Logic

If you are ready to configure your Input and Output Mapping, you can proceed immediately. Steps 2, 3, and 4 are blank and ready to configure. See Module Overview: Data Logic for guidance on configuring Input Mapping, Output Mapping, and Filter Records.

Option 2: Review the Blueprint in Operations > Blueprints

If you want to review the full Blueprint before configuring Data Logic, navigate to Operations > Blueprints. Locate your Blueprint in the list — Blueprints are displayed in aggregate across all CRM accounts and object types. To identify which CRM account a Blueprint belongs to, hover over its ID to see the source tooltip.

From the Blueprint list, you can:

• Preview the table data using the Table Preview panel below the list
• Edit the Blueprint name or description using the pencil icon
• Upload a revised version if you want to modify the rows before applying the Blueprint

  Note: Generated Blueprints are a starting point. Review the rows carefully — AI generates rows based on the field values it observed in your CRM at the time of analysis, but the rows may not cover every case or reflect every value in your data. You may want to add, edit, or remove rows before applying the Blueprint to your records.

After reviewing your generated Blueprint, you are ready to configure Data Logic to apply it to your CRM records. Here is what to consider before you begin:

Review the Output Columns

The Blueprint's output columns represent the fields that will be updated when a match is found. If any output columns reference fields that do not yet exist in your CRM, you will need to create those fields before you can map them in Data Logic. Output column suggestions from AI are meant to illustrate possibilities — you are not required to use all of them, and you can choose to map only the columns that are relevant to your use case.

Configure Input and Output Mapping in Data Logic

Return to Data Management > Data Logic. The generated Blueprint will be loaded under Step 1: Pick Blueprint. Configure the following:

• Step 2: Input Mapping — map the Blueprint's input columns to your CRM fields and select matching criteria
• Step 3: Output Mapping — map the Blueprint's output columns to your CRM fields and configure update conditions and fallback values
• Step 4: Filter Records — optionally configure filters to limit which records Data Logic evaluates

Run in Preview Mode First

Before applying changes to your CRM, run Data Logic in Preview mode to review which records would be updated and what values would be written. This is especially important with a newly generated Blueprint, since the rows may not cover every case in your data.

For full guidance on configuring and running Data Logic, see Module Overview: Data Logic.

Suggested for your data and Common use cases are two different paths for generating a Blueprint with AI, and they produce different results when you click Generate.

Suggested for your data analyzes your actual CRM field metadata — including field population, cardinality, and value distributions — and recommends Blueprint use cases based on what it finds in your specific data. For example, a suggestion for job title classification appears because your data has inconsistent job title variants that the AI detected. A suggestion for country normalization appears because your country field contains multiple spellings of the same value. Because the AI has access to your actual field values, it can generate a complete Blueprint immediately — the rows reflect real values from your data and are ready to review and apply.

Common use cases starts from business processes that Data Logic is well suited for, rather than from your data. The AI checks whether the fields typically needed for each use case exist in your CRM and presents relevant patterns accordingly. These cover multi-field business logic — territory assignment, lead routing, scoring, account health, compliance enforcement, and others. When you generate a Blueprint from a Common use case, the result is a starting point template with rows that reflect your data where possible, plus representative placeholder rows for patterns the AI inferred from the category. The output columns — Territory, Risk Level, and similar — are new columns you will map to your actual CRM fields when you configure Data Logic. You may also want to add, remove, or adjust rows to reflect your specific business logic before running.

Both paths lead to the same Data Logic configuration experience. The difference is how much of the Blueprint is ready to use versus how much you will customize before running it against your records.

The Data Logic Input and Output Mapping fields are blank after saving a generated Blueprint because Generate with AI only creates the Blueprint table. The mapping configuration is specific to your CRM fields and update conditions, which you manually set in Steps 2 and 3 of the Data Logic module. After saving, the Blueprint is automatically loaded into Step 1: Pick Blueprint in Data Logic, ready for you to configure. See Module Overview: Data Logic for guidance on configuring Input and Output Mapping.

Yes, a Blueprint can be regenerated if the results don't look right. On the Preview tab of the Generate with AI popup, click Back to return to the Guide AI tab. Edit the instructions in the Guide AI with your own words field to refine your logic, then click Preview again to generate a new Blueprint. You can iterate as many times as needed before clicking Save.

Output columns in a generated Blueprint may reference fields that don't yet exist in your CRM because they are suggestions meant to illustrate possibilities, not fields that are required to exist. You are not required to use all suggested output columns. Review the suggestions, decide which ones are relevant to your use case, create the corresponding CRM fields if needed, and then map only the columns you want to use in Step 3: Output Mapping.

Yes, generated Blueprint rows can be edited after saving. Navigate to Operations > Blueprints, locate and select your Blueprint, and click the Upload New Version icon in the Table Preview header. Download the current Blueprint CSV, make your edits, and upload the revised file as a new version. Insycle retains previous versions, allowing you to compare changes over time.

No, Generate with AI is not the only way to create a Blueprint. You can also create a Blueprint by uploading a CSV file — either using Create from CSV directly within Data Logic, or by navigating to Operations > Blueprints and clicking Upload. If you already have a Blueprint, you can select it in Data Logic using Select Existing, browse Insycle-provided examples using Explore Examples, or choose from Insycle-maintained reference datasets using Use Reference Data. For a full overview of all Blueprint sources, see Module Overview: Data Logic.

Yes, a generated Blueprint can be used as a starting point and edited manually — and this is often the recommended approach. Generate with AI creates a Blueprint based on the field values it observed in your CRM at the time of analysis, but the generated rows may not cover every case or reflect every value in your data.

After saving, navigate to Operations > Blueprints, locate your Blueprint, and click the Upload New Version icon in the Table Preview header to download the current CSV. Add, edit, or remove rows as needed, then upload the revised file as a new version. Insycle retains all previous versions, so you can compare changes over time.

Blueprints are designed to be maintained long-term. As your CRM data changes — new territories, updated scoring criteria, additional job title variants — you can upload new versions of the Blueprint to keep your logic current without reconfiguring your Data Logic templates. The templates continue to reference the same Blueprint and automatically use the latest version on their next run.

The same instructions will not always generate the same Blueprint. Generate with AI uses an AI model to generate Blueprint rows, and AI generation is non-deterministic — the same instructions may produce slightly different results each time, particularly in the specific values and row structure generated. If a generated Blueprint doesn't look right, return to the Guide AI tab, refine your instructions to be more specific, and regenerate.