Pro-Code Chunking of Data 360 Search Indexes for Agentforce

You build your vector database, craft an Agentforce agent, and link your search indexes to corporate knowledge. Then during user testing, your agent misses a glaring detail from a PDF or misinterprets a conversation entirely. The common reaction is to blame the LLM. The reality? Your chunking strategy fractured the data before the model ever saw it.

If a text snippet lacks proper context when retrieved from the vector database, your agent cannot generate an accurate response. This is the structural reality of retrieval-augmented generation (RAG): garbage in, garbage out.

This post shows how pro-code chunking for Data 360 search indexes using Data 360 Code Extension functions can be used to address three common chunking failures: fragmented data tables, multispeaker dialogue, and exposed personally identifiable information (PII). For each, it compares native chunking against custom pro-code chunking and shows the impact on agent responses.

Where native chunking falls short

Data 360 offers reliable native chunking for standard content. Mechanisms like Section-aware chunking and Semantic-based passage extraction cut documents along structural headings or thematic boundaries. They prevent sentences from splitting awkwardly and keep related paragraphs together.

However, enterprise data is rarely clean. It lives in complex multipage PDFs, regulatory records, dense tables, and multispeaker transcripts. When this data is fed into a character-count or token-length splitter, semantic continuity shatters.

How to write and deploy Code Extension functions

Note: The code examples in this post are intended to illustrate the concepts described in a simple way. They are designed for representative, happy-path examples, not for the full range of possible real-world inputs. Because the implementations rely on regular expressions (regex) and other pattern-based heuristics, they can produce both false positives and false negatives. For example, a heuristic might incorrectly classify an ordinary value as an account number or fail to recognize a valid table, speaker label, timestamp, or identifier in an unexpected format. Production-grade implementations of these functions are beyond the scope of this post.

Data 360 Code Extension functions let you deploy pro-code Python scripts natively within the Salesforce trust boundary for unstructured data pipelines where the built-in chunking behavior isn’t enough.

A function is a small, serverless-style unit that takes an input, performs a computation, and returns an output, running fully sandboxed and stateless. You author and debug it locally in your favorite IDE with the Data Custom Code SDK, deploy it through the Salesforce CLI or the UI, and wire it into your pipeline at the search index creation step. A code extension function can also call a generative or predictive model when your use case needs it. To go deeper, see the developer documentation.

How we tested

To test Code Extension custom chunking capabilities and compare them with native chunking capabilities, we evaluated three different documents. You can explore the structure of a chunking function and learn how to deploy it and run it during search index creation in the developer docs.

For each use case, we followed the same methodology:

1. Chunk the same source document via two separate paths:
   
   1. Path A (Native): The standard Agentforce Data Libraries (ADL) route with default chunking configuration
   2. Path B (Code Extension): Custom Python chunking function deployed via Code Extension
2. Query the resulting vector search indexes for Path A chunks using:
   
   1SELECT Chunk__c FROM vector_search(TABLE("index_name__dlm"), 'query', 
   2  'SourceRecordId__c="<adl-id>/<filename>"', 100)

3. Query the chunks for Path B by querying the unstructured data model object (UDMO) directly:
   
   1SELECT Chunk__c FROM vector_search(TABLE("index_name__dlm"), 'query', 
   2  'SourceRecordId__c="<adl-id>/<filename>"', 100)

4. Compare the chunks retrieved and evaluate what an agent would answer from each.

Use case 1: fragmented data tables

Imagine you are indexing a quarterly financial report (ABC_corp_q4_fy2025_financial_report.pdf): a 5-page PDF containing an executive summary, four dense multicolumn data tables, and forward-looking guidance.

Here are a few snapshots from the financial report document:

The Problem: Standard splitters often slice text strictly by character or token limits. Tables spanning multiple columns lose the relationship between headers and data rows. When your agent retrieves a chunk containing raw numbers, it has no column context to understand what those numbers mean.

You can solve this problem by implementing a structure-aware Python function that detects table boundaries (for example, markdown pipe tables, grid tables, and HTML tables), keeps each table as a unified block, and when a table exceeds the chunk size limit, splits it into sub-chunks with column headers repeated at the top of each sub-chunk.

Sample code

Native chunks: what you get without Code Extension

1SELECT Chunk__c FROM vector_search(TABLE("ADL_ABC_Financial_index__dlm"), 
2  'test something', 
3  'SourceRecordId__c="1JDKd000000kADxOAM/ABC_corp_q4_fy2025_financial_report.pdf"', 100)

Here is what native chunking produced, a wall of numbers with no structural context:

Native Chunk Example 1 (Raw Data Dump):

1DataCloud Pro 892 0.68 34.2 78.5 245.8 99.97 42 185 AgentForce 456 0.72 28.7 82.1 189.3 
299.95 38 162 Einstein Analytics 1,245 0.55 22.4 65.3 312.6 99.98 55 210 Flow Builder 678 
30.61 41.8 71.2 98.4 99.99 28 95 MuleSoft Connect 334 0.58 18.9 59.8 567.2 99.96 65 280 
4Tableau Cloud 1,567 0.45 26.3 68.9 78.5 99.94 120 450 Commerce Cloud 289 0.71 45.2 74.6 
5423.1 99.99 32 128 Marketing Cloud 1,102 0.52 31.5 72.8 198.4 99.96 48 195Service Cloud AI 
6567 0.66 25.8 69.4 156.2 99.97 35 142Slack Platform 2,340 0.78 52.1 85.2 892.4 99.99 22 
778 5. Forward-Looking Guidance - Q1 FY2026 Page 4/5

Native Chunk Example 2 (Fragmented Table):

1Enterprise Software 156.2 152.8 149.5 2.2 4.5 81.2 38.7 1,350 Professional Svcs 89.4 87.1 
282.6 2.6 8.2 42.1 15.3 980 Hardware & Infra 52.3 54.7 58.9 4.4 11.2 35.6 8.9 620 Emerging 
3Mkts & IoT 24.8 21.2 15.4 17.0 61.0 55.4 5.2 340 Security & Compliance 13.4 11.9 9.8 12.6 
436.7 78.1 22.4 210 AI & Automation 42.1 35.8 22.4 17.6 87.9 82.5 28.1 410 Integration 
5Platform 38.9 36.2 33.1 7.5 17.5 71.8 26.9 380Developer Tools 28.6 25.4 19.8 12.6 44.4 
676.3 18.2 290 Industry Solutions 52.8 48.9 41.2 8.0 28.2 69.4 24.5 450 Page 2/5

The Problem: As just one example, the numbers following “AI & Automation” (42.1 35.8 22.4 17.6 87.9 82.5 28.1 410) sit in a chunk with no column headers. What is 42.1? Revenue? A margin? A growth rate? Without the header row (Segment | Q4 FY25 Rev($M) | Q3 FY25 Rev($M) | ... | Headcount), these numbers are meaningless. Your agent cannot answer even basic questions about this data.

Code Extension chunks: structure preserved

The custom function detects table boundaries, preserves the header row, and produces structured chunks:

1SELECT Chunk__c FROM ADL_ABC_Financial_MultiPage_chunk__dlm

Custom Chunk Example 1 (Executive Summary — Clean Prose):

1ABC Corp - Q4 FY2025 Quarterly Financial Report (Confidential)
2
3**Executive Summary**
4
5Total revenue for Q4 FY2025 reached $847.3M, representing a 12% year-over-year increase 
6driven primarily by growth in our Cloud Services and Data Platform segments. Operating margin 
7improved to 28.4%, up from 25.1% in the prior quarter, reflecting improved cost discipline 
8and scale efficiencies across all business units.
9
10Key highlights include the successful launch of three new enterprise products, expansion into 
11the APAC market with two new regional data centers, and a 15% reduction in customer churn 
12attributed to our enhanced support model. The Data Platform segment showed particularly strong 
13momentum with 18.1% YoY growth, driven by increased adoption of our AI-powered analytics features.
14
15Free cash flow generation remained robust at $198.4M for the quarter, enabling continued 
16investment in R&D while returning $85M to shareholders through buybacks. Our balance sheet 
17remains strong with $2.1B in cash and short-term investments and a net-debt-to-EBITDA ratio of 0.8x.

Custom Chunk Example 2 (Table with Headers Preserved — Revenue by Segment):

1**1. Revenue Breakdown by Business Segment**
2
3The following table summarizes revenue performance across all business segments for Q4 FY2025, 
4with year-over-year and quarter-over-quarter comparisons, gross margin, and headcount allocation:
5
6* **Cloud Services**
7* The Segment has Q4 FY25 Rev($M) of 312.5.
8* The Segment has Q3 FY25 Rev($M) of 289.1.
9* The Segment has Q4 FY24 Rev($M) of 261.8.
10* The Segment has QoQ Growth% of 8.1.
11* The Segment has YoY Growth% of 19.4.
12* The Segment has Gross Margin% of 72.3.
13* The Segment has Op Margin% of 34.2.
14* The Segment has Headcount of 2,450.
15
16* **Data Platform**
17* The Segment has Q4 FY25 Rev($M) of 198.7.
18* The Segment has Q3 FY25 Rev($M) of 185.3.
19* The Segment has Q4 FY24 Rev($M) of 168.2.
20* The Segment has QoQ Growth% of 7.2.
21* The Segment has YoY Growth% of 18.1.
22* The Segment has Gross Margin% of 68.9.
23* The Segment has Op Margin% of 31.5.
24* The Segment has Headcount of 1,820.

Custom Chunk Example 3 (Regional Performance — Each Region Self-Contained):

1**2. Regional Performance Matrix**
2
3Geographic revenue distribution across all operating regions, including customer metrics, 
4deal economics, and satisfaction indicators:
5
6* **APAC - ANZ**
7* Revenue ($M): 52.8
8* % of Total: 6.2%
9* YoY Growth %: 18.4%
10* Customers: 486
11* Avg Deal ($K): 108.6
12* NPS: 67
13* Renewal %: 90.2%
14* Expansion %: 114.6%
15
16* **APAC - Japan**
17* Revenue ($M): 34.2
18* % of Total: 4.0%
19* YoY Growth %: 21.5%
20* Customers: 312
21* Avg Deal ($K): 109.6
22* NPS: 71
23* Renewal %: 91.8%
24* Expansion %: 116.2%

Impact on agent responses

User Query: “What is the YoY growth rate for the AI & Automation segment and how does its operating margin compare to Cloud Services?”

Use case 2: multispeaker dialogue

Now consider a use case in which you’re grounding your service agents on a recorded customer escalation call: a 45-minute platform migration planning session with five participants (Sarah Chen, VP of Engineering; Marcus Webb, Solutions Architect; Priya Desai, Database Lead; Tom Nakamura, DevOps Manager; Lisa Okafor, Customer Success Director).

Source document: Globex_Migration_Escalation_Call_2025_Q4.pdf

Here is a snapshot from the source document:

The Problem: Standard chunking produces massive walls of unstructured text. Speaker boundaries are lost, timestamps disappear, and topical shifts get merged into a single chunk. The result: chunks thousands of characters long where you cannot tell who said what or when.

You can solve this by using a windowed sliding dialogue mechanism with these parameters: WINDOW_SIZE = 6 turns, STRIDE = 3 turns, MAX_CHUNK_CHARS = 1500. This helps ensure that every question and its answer are almost always co-located in at least one chunk. Adjacent chunks overlap so context is virtually never orphaned.

Sample code

Native chunks: what you get without Code Extension

1SELECT Chunk__c FROM vector_search(TABLE("ADL_MultiSpeakerCon_index__dlm"), 
2  'test something', 
3  'SourceRecordId__c="<library-id>/Globex_Migration_Escalation_Call_2025_Q4.pdf"', 100)

Native Chunk Example (Wall of Text — No Speaker Attribution):

1Marcus Webb: I think yeah I th I think if you can just get a number whether it's through 
2the benchmarks we ran last month or whether we have to do new load tests. If you can just 
3get a number and saying if you had to migrate the legacy Oracle instances to the new 
4Postgres cluster, how much data volume can we migrate within that within 72 hours or I 
5guess even conservatively within 48 hours. Let's go with 48 hours because we probably 
6need a buffer for rollback as well. Um do you know just just coming up with numbers of 
7how much data we can migrate within a period of timewill give us some kind of uh 
8measurement to know how many tenant databases we can cut over within that window or we 
9can commit to Sarah Chen: Yeah,Marcus Webb: enough Sarah Chen: because the the challenge 
10is like uh we'll be penalized right if we commit to uh certain migration windows if 
11we're not able to meet that. 00:25:46 Sarah Chen: So this is a very crucial part of it 
12wherein uh we have to kind of come up with these numbers based on uh the actual load 
13testing uh given that we're not going the blue-green deployment route and we're doing it 
14on a rolling basis. So I think we need that sort of analysis to be Marcus Webb: Priya,
15Tom Nakamura: A Marcus Webb: is that something Kevin can can work with the DBA team 
16Sarah Chen: done. Tom Nakamura: minute. Marcus Webb: on? Priya Desai: Yeah, I'll follow 
17up with Kevin after this Marcus Webb: Okay, thanks.

The Problems:

1. No speaker separation: Turns from Marcus, Sarah, Tom, and Priya are mashed together in a single paragraph
2. Timestamps embedded randomly: 00:25:46 appears midsentence with no structural break
3. Questions divorced from answers: Marcus asks “can Kevin work with the DBA team?” and Priya’s answer is buried without clear attribution
4. Topic bleed: The Oracle-to-Postgres migration capacity discussion runs directly into the SLA penalty discussion with no boundary

Code Extension chunks: windowed dialogue

The custom function parses speaker turns, then creates overlapping windows:

Custom Chunk Example 1 (Clear Speaker Attribution and Overlap):

1Sarah Chen: uh so I would uh so that's what my categorization of phases and milestones is. 
2Phase one is more longer term more strategic where we want to land up eventually right we do 
3want to uh achieve uh reach a point wherein for all the enterprise tier customers we want to 
4uh have a zero-downtime migration kind of an approach uh but you know in terms of milestones 
5now milestone is more like uh you know various different checkpoints that we have so our first 
6milestone is towards end of Q4 when we want to
7
8Sarah Chen: in Yeah. Yeah. So basically that was the next thing that I was going to talk 
9about. If you see this these bottom uh you know tiers. So for all the enterprise tier 
10customers what we offer them is guaranteed migration within the committed downtime windows 
11you know uh they pay for a contractual guarantee. Whereas for all the customers who don't uh 
12uh purchase the premium migration package, it'll be on a best effort basis without any 
13committed timelines, no contractual SLAs, but they'll
14
15Sarah Chen: anywhere uh we have like three layers. The first two layers talks about schema 
16migration and data replication setup. We know the CDC streams are running every 5 minutes. 
17The full table snapshots you know every 12 hours and we are also trying to you know make uh 
18target uh to reduce the 12-hour sync window to 2 hours for Q4 and uh the development is 
19complete and we are in the process of you know testing it and the third layer is where uh you 
20know the the actual cutover needs to happen

Custom Chunk Example 2 (Decision and Response Together):

1Tom Nakamura: correct as long as uh we are meeting their downtime window requirement they 
2don't bother how we are providing Sarah Chen: might Tom Nakamura: them that migration how 
3efficient
4
5Sarah Chen: uh so I would uh so that's what my categorization of phases and milestones is. 
6Phase one is more longer term more strategic where we want to land up eventually right we do 
7want to uh achieve
8
9Sarah Chen: in Yeah. Yeah. So basically that was the next thing that I was going to talk 
10about. If you see this these bottom uh you know tiers. So for all the enterprise tier 
11customers what we offer them is guaranteed migration within the committed downtime windows
12
13Sarah Chen: anywhere uh we have like three layers. The first two layers talks about schema 
14migration and data replication setup.
15
1600:55:02
17
18Sarah Chen: Yeah. Yeah. We Tom Nakamura: Hey. Uh, so our milestone it talks about in terms 
19of per tenant for Q4,

Key Difference: Each chunk now contains multiple turns from the same topical exchange. The sliding window (stride of 3) helps ensure that a question asked in one chunk is always paired with its answer. Speaker names are preserved as structural anchors.

Impact on agent responses

User Query: “What is the committed downtime window for enterprise tier customers during the database migration?”

Use case 3: PII masking and regulatory cleansing

In this use case, you’re operating in a regulated sector (for example, healthcare, banking, or public sector) where internal identifiers must be sanitized before ingestion into a vector database that agents will query.

Source document: Sample_PII_Masking_Test_Document.pdf is a five-page document with simulated clinical intake records, banking credit memos, and public sector benefits cases. It contains both standard PII and proprietary identifiers.

Here is a snapshot from the source document:

The Problem: Standard pattern recognition is not applied during chunking via the ADL route. All identifiers, standard and proprietary, pass through into the vector index unmasked.

You can address this by implementing regex-based pattern matching over the raw text inside your custom function before chunking. This cleanses SSNs, credit card numbers, phone numbers, email addresses, employer identification numbers (EINs), ICD codes, and other industry-specific formats on-the-fly. The data is sanitized before it ever reaches the chunk data model object (DMO).

Sample code

Native chunks: all PII exposed

1SELECT Chunk__c FROM vector_search(TABLE("ADL_PIIMasking_index__dlm"), 
2  'test something', 
3  'SourceRecordId__c="1JDKd000000kAE7OAM/Sample_PII_Masking_Test_Document.pdf"', 100)

Native Chunk Example 1 (Healthcare — Full PII Exposed):

Note: The X characters below are synthetic placeholders; they represent identifiers that are present and unmasked, not values that have already been redacted.

1CONFIDENTIAL - INTERNAL USE ONLY Meridian Regional Health System Patient Intake & Clinical 
2Summary - Q4 2025 Patient Demographics Standard Identifiers Patient Name: Margaret A. Thornton 
3Date of Birth: 01/01/1970 Social Security Number: XXX-XX-XXXX Phone: (XXX) XXX-XXXX Email: 
4m.thornton@example.com Address: 123 Main St, Anytown, IL 00000 Proprietary & 
5Regional Identifiers MRN (Medical Record No.): MRH-XXXX-XXXXXX-X Internal Episode ID: 
6EP.Q4.NAP.093871.TNT State HIE Patient Token: ILHIE-PKT-XXXXXXXX-XXXX Payer Member ID: 
7BCBSIL-GRPXXXX-MXXXXXXX Referring Provider NPI+Loc: NPI1629384750-LOC.NAP.03 Bed Assignment 
8Code: 4N-BED.217-ISO-RESP Regional Trauma Registry ID: ITRS-R5-2025-00847

Native Chunk Example 2 (Banking — All Identifiers Visible):

1Credit Analysis Narrative Subject: Apex Manufacturing Solutions (CIF: CNB-CIF-2025-093871-COMM) 
2requests $4.2M term loan expansion. Current exposure tracked under facility ID 
3FAC-CNB-MW-2025-APEX-001. Borrower maintains primary operating account 
4(DDA-CNB-MW-7741093871) with average collected balance of $847K. Guarantor Robert J. Whitfield 
5(Internal KYC Profile: KYC-IND-XX-XXXX-XXXXXX, SSN:XXX-XX-XXXX) provides unlimited 
6assessment net worth WA-CNB-2025-0291-WHTFLD. personalguarantee. Guarantor verified via 
7internal wealth Collateral: IL. Appraisal ordered under engagement APR-CNB-2025-093871-IND 
8(Appraiser Vendor ID: VND-APR-MW-0087-CERT). Environmental Phase I completed (Report ID: 
9ENV-PH1-2025-APEX-093871).

Native Chunk Example 3 (Cross-System — Re-identification Risk):

1Event 2: Employment Verification Timestamp: 2025-11-03T09:15:00Z Source: IL Dept of Revenue 
2(System ID: SYS-ILDOR-EMPL-VERIFY) Destination: DHS Benefits (System ID: 
3SYS-DHS-IL-ELIG-ENGINE) Subject: SSN XXX-XX-XXXX / DHS-IL-XXXX-XXXX-XXXXXX-XX Employer Match: 
4ILDOR-EMP-2025-00-0000000-APEX Income Verified: 
53,847.00/month(PayPeriodID:PP-APEX-2025-B22) Verification Token: ILDOR-VT-2025-1103-093871-CONF

The Problem: Every sensitive identifier sits unmasked in the vector index:

• SSNs: XXX-XX-XXXX, XXX-XX-XXXX, XXX-XX-XXXX
• Credit card number: XXXX-XXXX-XXXX-XXXX
• Email address: m.thornton@example.com
• Phone number: (XXX) XXX-XXXX
• Medical record number: MRH-XXXX-XXXXXX-X
• Internal KYC profile: KYC-IND-XX-XXXX-XXXXXX
• Payer member ID: BCBSIL-GRPXXXX-MXXXXXXX
• DHS case number: DHS-IL-XXXX-XXXX-XXXXXX-XX

When your agent retrieves these chunks, it may echo PII directly back to end users, which constitutes a compliance violation under HIPAA, PCI-DSS, and many state privacy laws.

Code Extension chunks: PII masked before indexing

The custom function applies regex-based pattern matching to sanitize sensitive data before chunking:

Custom Chunk Example 1 (Healthcare — Standard PII Redacted):

1CONFIDENTIAL - [SWIFT_REDACTED] USE ONLY Meridian Regional Health System Patient Intake & 
2Clinical Summary - Q4 2025 Patient Demographics Standard Identifiers Patient Name: Margaret A. 
3Thornton Date of Birth: 01/01/1970 Social Security Number: [SSN_REDACTED] Phone: 
4([PHONE_REDACTED] Email: [EMAIL_REDACTED] Address: 123 Main St, Anytown, IL 00000 
5Proprietary & Regional Identifiers MRN (Medical Record No.): MRH-XXXX-XXXXXX-X Internal 
6Episode ID: EP.Q4.NAP.093871.TNT

Note the ‘[SWIFT_REDACTED]’ token replacing ‘INTERNAL’: an over-broad SWIFT-code pattern matched a plain word. Also note that certain PII,  such as names, addresses, dates of birth are not handled by the example function).

Custom Chunk Example 2 (Banking — Standard PII and Financial IDs Redacted):

1Primary Contact SSN: [SSN_REDACTED] DUNS Number: 07-481-2937 Credit Card on File: 
2[CREDIT_CARD_REDACTED] (Exp: 09/27) Internal & Proprietary Identifiers Relationship Manager 
3ID: RM-COMM.MW.4471.SR Internal Credit Score: CNB-ICS-782-TIER2A Collateral Tracking ID: 
4COLL-RE.IL.2025.093871.INDUST Loan Pipeline Position: PL-Q4.2025-MW-00291-PEND Risk Rating 
5(Proprietary): RR-CNB-3B-WATCH-MOD BSA/AML Case Reference: AML-SAR.2025.MW.04471 Treasury 
6Mgmt Client Code: TM-APEX-MW-4471-SWEEP

Custom Chunk Example 3 (Cross-System Events — SSN and EIN Masked):

1Event 2: Employment Verification Timestamp: 2025-11-03T09:15:00Z Source: IL Dept of Revenue 
2(System ID: SYS-ILDOR-EMPL-VERIFY) Destination: DHS Benefits (System ID: 
3SYS-DHS-IL-ELIG-ENGINE) Subject: SSN [SSN_REDACTED] / DHS-IL-XXXX-XXXX-XXXXXX-XX Employer 
4Match: ILDOR-EMP-2025-[EIN_REDACTED]-APEX Income Verified: $3,847.00/month (Pay Period ID: 
5PP-APEX-2025-B22) Verification Token: ILDOR-VT-2025-1103-093871-CONF

Custom Chunk Example 4 (Public Sector — Government IDs Masked):

1CONFIDENTIAL - [SWIFT_REDACTED] USE ONLY State of Illinois - Department of Human Services 
2Benefits Determination & Case Management Summary Claimant Information Standard Identifiers 
3Claimant Name: David R. Okonkwo Social Security Number: [SSN_REDACTED] Date of Birth: 
401/01/1980 Driver's License: O-000-0000-0000 (IL) Phone: ([PHONE_REDACTED] Agency-Specific 
5& Regional Identifiers DHS Case Number: DHS-IL-XXXX-XXXX-XXXXXX-XX

What got caught and what needs custom patterns

In this test, the Code Extension function successfully redacted:

• SSNs: XXX-XX-XXXX → [SSN_REDACTED]
• Credit card numbers: XXXX-XXXX-XXXX-XXXX → [CREDIT_CARD_REDACTED]
• Email addresses: m.thornton@example.com → [EMAIL_REDACTED]
• Phone numbers: (XXX) XXX-XXXX → [PHONE_REDACTED]
• EINs: 00-0000000 → [EIN_REDACTED]
• ICD Codes: J96.01 → [ICD_CODE_REDACTED]

Proprietary identifiers that require organization-specific patterns (these passed through, demonstrating why you must extend the pattern registry for your formats):

• MRH-XXXX-XXXXXX-X — Medical record number
• BCBSIL-GRPXXXX-MXXXXXXX — Payer member ID
• DHS-IL-XXXX-XXXX-XXXXXX-XX — DHS case number
• ILHIE-PKT-XXXXXXXX-XXXX — HIE patient token

The key insight: Code Extension gives you the framework to add any pattern your organization needs. Standard numeric identifier patterns are handled by the existing sample code. Proprietary formats are added as one-line regex entries:

code here 22

Impact on agent responses

User Query: “What was the clinical pathway for the patient admitted on October 17th?”

Compliance implications

The bottom line: without Code Extension, everything goes into the vector store unmasked. With the Code Extension function, PII formats covered by the sample patterns are masked  during search index chunking, and the extensible pattern registry gives you a path to add proprietary formats as you discover them.

Conclusion

Code Extension functions are not a replacement for native chunking; they provide an escalation path when enterprise data complexity outpaces standard algorithms. Investing in pro-code chunking ensures your search indexes remain contextually rich and drive accurate Agentforce responses.

Get started with Code Extension

What are you waiting for? Start building Code Extensions today: Enable Code Extension in Data 360 setup feature manager, install the Data Custom Code SDK and the Salesforce CLI, and start building right away. To learn more:

• Check out the Code Extension function documentation.
• Watch the Extend Data 360 with Code Extension YouTube playlist.

About the authors

Jitin Mehndiratta is a Product Management Director at Salesforce working on Data 360. He focuses on enabling developers to build intelligent data pipelines that power Agentforce experiences.

Samaritha Patlori is a Product Manager for Code Extension at Salesforce Data 360. With 7+ years across engineering and product, she is currently focused on enabling developers to run custom code securely within Data 360.