Open CTI Retirement Guide: Moving Salesforce Contact Centers Toward Agentforce

Guest blog by: Antonina Kharchenko

The countdown has officially begun for legacy telephony integrations. With February 28, 2028, marked as the Salesforce Open CTI end of life, IT teams and customer experience directors face a critical juncture. This isn't just another routine software patch or minor version update; it is a structural mandate to modernize how customer interactions are handled, routed, and resolved.

If you track recent call center AI news, the driving force behind this shift is undeniable. Industry forecasts anticipate that artificial intelligence will manage up to 50% of routine customer service inquiries by 2027. Clinging to outdated, disconnected frameworks ensures your operations will lag behind this massive wave of automation. The Open CTI retirement is an opportunity to rebuild a faster, smarter communication hub.

Here is your technical and strategic playbook for navigating the transition.

Diagnosing the Legacy Bottleneck

To map out a successful migration, we first need to understand why the existing framework is being phased out. For years, Open CTI served as a highly effective, browser-based bridge. It connected external telephony providers, such as Avaya and Cisco, and local PBX systems directly to the CRM without requiring clunky desktop software installations.

However, as the Salesforce Open CTI ends, its architectural flaws are holding businesses back. The core problem is data fragmentation. In an Open CTI setup, the external telecom provider handles all the heavy lifting: the actual audio stream, call recording, IVR menus, and routing logic. Salesforce merely receives a ping containing metadata (such as caller ID or call duration).

Because the systems are fundamentally decoupled, achieving real-time intelligence is nearly impossible. Organizations are forced to maintain custom JavaScript for every unique vendor API, leading to a fragmented user interface for agents and high technical debt for developers.

The Standard: A Native, Intelligent Engine

The replacement strategy revolves around consolidating operations into a unified Salesforce AI contact center. The combination of Salesforce Voice and Agentforce represents this new baseline.

Instead of relying on a brittle external bridge, Salesforce Voice brings the telephony experience natively into the platform. Calls are answered, transcribed, and logged inside the CRM. This unified data pool is exactly what Agentforce needs to operate effectively.

When you transition to an AI-powered contact center Salesforce model, artificial intelligence stops being a post-call analytics tool and becomes a real-time participant. During a live interaction, the AI monitors the conversation, gauges customer sentiment, automatically retrieves relevant knowledge base articles, and prompts the agent with next-best actions.

The operational impact is highly measurable:

• 93% of service professionals report that integrated AI directly saves them time.
• Agents actively utilizing these tools reduce time spent on routine, repetitive tasks by 20%.
• Salesforce reports that Agentforce has already managed over 2 million conversations via Salesforce Help.

Voice call record page with transcription and next-best-action recommendation

The Enterprise Solution: A Hybrid Voice Architecture

While a fully native Salesforce voice AI setup is ideal for some, it is not a realistic immediate step for massive, complex enterprises. Many organizations operate with multi-regional hardware deployments, strict data residency compliance laws, or multi-year contracts with major telecom carriers. They cannot simply abandon their existing telephony infrastructure overnight.

For these complex environments, the solution is implementing an Enterprise Voice Control Layer.

This middleware approach allows a "Bring Your Own Telephony" (BYOT) strategy. Specialized applications found on the AppExchange act as an intelligent orchestrator. They allow you to maintain your current Avaya, Cisco, or Microsoft Teams routing for voice delivery, while seamlessly pushing the real-time interaction data into Salesforce’s AI engine.

By leveraging tools like AMC Technology's DaVinci, companies can trigger background identity authentication and feed live audio streams into Agentforce for real-time transcription, all without ripping out their underlying telecom hardware. Considering 76% of consumers now expect highly personalized and immediate service, deploying this intelligent middle layer ensures agents have the context they need the second the call connects.

Enterprise voice control layer solutions on AppExchange

Architectural Comparison Breakdown

Understanding your deployment options is critical. Here is how the three main architectural paths compare:

The Migration Playbook

Successfully moving away from legacy adapters requires a structured, phased approach to avoid dropping calls or losing data. Follow this sequence to safeguard your operations:

1. Conduct a Deep Infrastructure Audit: Document every active CTI adapter, custom workflow, and screen-pop rule currently running in your environment. Identify exactly what data is being passed from your telecom provider to the CRM.
2. Determine Your Architectural Path: Based on the audit, decide if your organization can migrate fully to a native Salesforce environment, or if you need an enterprise orchestration layer to protect existing vendor investments.
3. Deploy the Connectivity Layer: If utilizing a hybrid setup, install and configure your central integration framework. Establish secure connections between your existing voice hardware and the Salesforce intelligence engine.
4. Pilot the Intelligence: Roll out live transcription and Agentforce sentiment analysis to a small, controlled group of service reps. Use their feedback to fine-tune automated workflows and routing rules.
5. Execute a Parallel Run: Run the legacy Open CTI setup alongside your new architecture for a designated testing period. Once stability and data accuracy are confirmed, scale the new solution company-wide and decommission the old adapters.

Conclusion

The retirement of Open CTI mandates a permanent shift in customer service infrastructure. Relying on fragmented, delayed call data is no longer a viable operational strategy. By planning your migration pathway now, whether through a fully native environment or a strategic enterprise voice layer your contact center will be positioned to leverage real-time intelligence, cut resolution times, and meet the rising expectations of today's consumers.

Salesforce Data Quality: How to Audit Your Org Before AI and Automation

Guest blog by: Mykhailo Radchenko

Artificial Intelligence is no longer a futuristic concept; it is an active component of daily business operations. Many companies are currently in a race to integrate AI into their workflows, investing heavily in intelligent automation to achieve significant productivity gains. However, a silent hurdle is preventing these investments from reaching their full potential: messy data.

The Financial Reality of Poor Data Quality

The impact of bad data is cumulative and affects every corner of an organization. To prevent marketing campaigns from missing targets and forecasting from becoming unreliable, organizations often rely on a Salesforce data quality playbook to standardize their data entry and maintenance processes. This translates into significant financial losses. According to a 2024 Forrester Research report, over 25% of global data and analytics employees estimate annual losses exceeding $5 million due to poor data quality, with 7% reporting losses of $25 million or more. Gartner further supports this, noting that poor data quality costs organizations an average of $12.9 million annually.

Perhaps the most pressing concern is the threat to future competitiveness. Gartner predicts that through 2026, organizations will abandon 60% of AI projects that lack AI-ready data. This unreliability also severely impacts customer retention. According to a Zendesk study, over 50% of consumers will switch to a competitor after a single bad experience, and 73% after multiple poor experiences. 

Image source: Gartner

Why Companies Struggle with Salesforce Hygiene

For many firms, Salesforce acts as the "central nervous system" of their go-to-market engine. But as companies grow, they build complex technology stacks where marketing automation, ERP systems, and sales platforms all feed data into the CRM. These multiple entry points often result in a CRM filled with missing pieces and mixed-up details.

According to a recent study by Salesforce, the vast majority of executives–87%–view data silos as the primary hurdle preventing them from using artificial intelligence effectively. Despite this, a Salesforce/Forrester survey found that two out of three companies do not have a proper data strategy, although many of them already use AI.

Image source: Salesforce

Furthermore, data decay happens faster than most teams realize. At least 28% of business email addresses expire within a single year, according to a recent industry report. This means that without a consistent strategy to maintain data quality, more than a quarter of your database could be obsolete within 12 months. When Salesforce data hygiene is neglected, the results include:

• Inaccurate Forecasting: 39% of sales professionals claim poor data prevents accurate pipeline reporting.
• Lost Productivity: Sales reps spend 70% of their time on non-selling tasks, a figure largely unchanged since 2022.
• AI Failure: 63% of sales professionals report that their company’s data is not properly set up for generative AI.
• Loss of Trust: 65% of sales professionals report they cannot fully trust their organization's data.

Understanding Salesforce AI Data Quality Dimensions

Are you ready for AI? The growth potential is significant: Salesforce research indicates that 90% of SMB leaders report AI makes operations more efficient, while 87% say it helps scale services, and 86% believe it improves margins and competitive standing. Furthermore, sales teams that deploy AI with reliable foundations see a clear revenue advantage: 83% saw gains, compared to only 66% of teams without AI. However, these results are achievable only if the underlying data are reliable.

Image source: Salesforce

Before deploying bots or predictive models, it is essential to understand the data quality dimensions Salesforce AI requires to function. If the underlying data is fragmented or incomplete, AI tools will likely surface outdated info or produce "hallucinations". In one high-profile case, an AI support bot invented a fake login policy and sent it to users without human oversight, leading to canceled subscriptions and a public apology. To improve the quality of the data in Salesforce, organizations should audit their org against these four key dimensions:

1. Required Field Standards

AI models require specific "ingredients" to produce meaningful results. For objects like Leads, Contacts, and Opportunities, you must identify which fields, such as Industry, Job Title, or Annual Revenue, are mandatory for your specific AI use case. Identifying these gaps is the first step in creating a reliable dataset.

2. Picklist Uniformity

Inconsistent values are a primary cause of broken logic in AI segmentation. If your "Industry" field contains variations such as "Healthcare," "Health Care," and "Medical," an AI will treat them as separate categories. Normalizing these picklists is vital for Salesforce data quality.

3. Duplication Thresholds

You must define what constitutes a duplicate within your specific business context. For example, should "IBM" and "International Business Machines" be merged?. Using Salesforce deduplication filters to define these match criteria is necessary to prevent the AI from processing redundant or conflicting information.

4. Record Freshness

The "New/Modified Records" trend view is a useful metric to identify stale data. Tracking records created or updated over the past 360 days helps flag data that might confuse analytics or cause poor customer experiences.

The "New/Modified Records" trend view is a useful metric to identify stale data. Tracking records created or updated over the past 360 days helps flag data that might confuse analytics or cause poor customer experiences.

How to Improve Data Quality in Salesforce: A Phased Approach

Cleaning an entire Salesforce instance can feel overwhelming, but a phased strategy for Salesforce data cleansing can yield immediate results.

Phase 1

Target Exact Matches: Start with the "low-hanging fruit" – obvious exact matches where Leads or Contacts share the same email and account name. Merging these records immediately restores trust in the CRM for the teams using it every day.

Phase 2

Automate and Schedule: Once the initial "mess" is cleared, schedule automated merge jobs to handle low-risk duplicates in the background. This ensures that your efforts to maintain data quality in Salesforce are consistent and not just a one-time project.

Phase 3

Prevent at the Source: The most effective way to improve Salesforce data quality is to stop errors before they enter the system. Using API integrations, such as those offered by Cloudingo, allows external systems, such as marketing automation or ERPs, to be deduplicated before they land in Salesforce.

Real-World ROI: The Cost of Inaction

The need for a Salesforce data quality playbook is often best illustrated by companies that have faced "data disasters". For instance, Docker, Inc. faced a situation in which thousands of duplicate records filled its system – some companies appeared as 60 separate accounts due to repeated credit card payments. This led sales reps to stop trusting the CRM entirely, resulting in nearly $1 million in annual lost revenue.

Similarly, at 1-800Accountant, the team was "drowning in duplicates," with over 32,000 bad leads. This led to one in five sales calls being repeats, with clients contacted up to 10 times, resulting in significant prospect frustration. 

A similar situation occurred at Lucid Design Group, where a CRM with 75,000 records became so unmanageable that the sales team repeatedly called disqualified leads, and the sole Admin spent a full working day manually merging duplicates.

By implementing a structured audit and an automated cleanup solution, these companies restored trust and provided their teams with a reliable system.

Conclusion: Building the Future on a Clean Foundation

Salesforce data quality is no longer just a background administrative task; it is a core growth strategy. As organizations prepare for deeper AI integration, they must move away from manual, reactive cleanup and toward proactive, automated maintenance.

If your data is messy, AI will only amplify those errors. But if your data is clean, AI becomes a powerful multiplier for productivity and revenue. For a deeper dive into these strategies, you can download the full Salesforce data quality playbook or explore the technical requirements for Salesforce data hygiene.

Designing Salesforce Payment Workflows with QuickBooks and Stripe (Part: 2)

Guest blog by: Antonina Kharchenko

Introduction: Operating Salesforce Payment Workflows in Real-World Integrations

In the first part of this guide, we explored how to design Salesforce payment workflows when integrating with systems such as Stripe and QuickBooks. We discussed how to structure Salesforce objects, define workflow triggers, model integration fields, and automate billing‑related processes so that customer records, subscriptions, and invoices remain aligned across platforms. If you haven’t read it yet, you can start with our previous article.

While designing workflows is essential, running payment integrations in production introduces a different set of challenges. Once Salesforce begins exchanging data with external financial systems, teams must manage synchronization timing, API limits, record matching logic, and security considerations.

These challenges are not unique to payment systems, they reflect broader trends in enterprise technology. According to Salesforce and MuleSoft’s Connectivity Benchmark Report, organizations today use over 1,000 applications on average, yet less than 30 % of these applications are integrated, leaving vast portions of business data siloed and disconnected.

Average Apps per Organization vs. Integrated Apps, image from Salesforce.

This article focuses on the most common challenges teams encounter when running Salesforce payment workflows in real environments. Understanding these challenges helps teams anticipate problems early and build integrations that remain stable as systems and data volumes grow.

Common Challenges in Salesforce Payment Workflows

Different challenges can come up depending on how you integrate Salesforce with external systems. Most issues happen with custom-built integrations, while some can also appear when using middleware or prebuilt apps. The challenges in this section give you a clear idea of what to watch out for.

Issue #1. Record Matching and Duplicate Accounts

What happens:

QuickBooks and Stripe both maintain their own unique customer IDs and account structures. When Salesforce attempts to sync a client record, the integration must match the Salesforce record to an existing external record.

Typical failure modes:

• The external system creates a new customer instead of matching an existing one.
• Duplicates are introduced because the matching criteria differ (email vs. name vs. ID).

What to check:

• How the integration matches records by default (email, external ID, custom logic).
• Whether custom matching rules can be configured.

Why it matters: Duplicate clients lead to mismatched financial records and require manual cleanup. A Salesforce study found that the average customer’s contact database is composed of 90% incomplete contacts, with 20% of records being useless due to several factors, such as 74% of the records needing updates and more than 25% of those being duplicates.

Issue #2. Field Mapping Discrepancies Between Systems

What happens:

Salesforce objects do not map one‑to‑one with external system objects. QuickBooks Online uses nested address structures (e.g., BillAddr, ShipAddr) and requires a CustomerRef when creating invoices. When updating records, SyncToken must be provided to prevent conflicts. Stripe also has its own object schema with required fields for customers and subscriptions. 

Typical failure modes:

• Missing required fields cause API errors.
• Updates fail if required attributes like SyncToken are not supplied.

What to check:

• Confirm mandatory fields in QuickBooks or Stripe before creating or updating records.
• Verify that Salesforce fields are mapped to the correct external attributes.

Issue #3. Timestamp and Data Staleness

What happens:

Different systems update data at different times and may use independent clocks or refresh intervals. Salesforce may receive external data with a delay or with timestamps that do not exactly match internal expectations.

Typical failure modes:

• A subscription change in Stripe is reflected in Salesforce much later.
• QuickBooks invoice states update with a delay and appears outdated.

What to check:

• Sync frequency (near‑real‑time vs. scheduled batch).
• Whether the integration supports event‑driven updates or only polling.

For example, with Breadwinner, you can configure the sync schedule to control how often updates are pulled into Salesforce.

Issue #4. API Limits and Large Data Loads

What happens:

Stripe and QuickBooks APIs impose rate limits. When initial data loads are large (many customers, invoices, payments), or when webhook traffic spikes, calls can fail or get throttled.

Typical failure modes:

• Integration stops due to rate limit rejections.
• Partial imports without clear error diagnostics.

What to check:

• API usage quotas for the external system.
• Support for throttling and back‑off logic.

Issue #5. Integration Security and Least Privilege

What happens:

Integration connections need the right permissions to create, update, or delete records in Salesforce, Stripe, or QuickBooks. If an integration user has too many privileges, it increases security and compliance risk; if it has too few, API calls will fail. Storing credentials insecurely (like hard‑coding in config or metadata) also poses a risk.

Typical failure modes:

• API calls fail due to insufficient permissions.
• Credentials embedded in code or configuration create audit and security issues.
• Over‑privileged integration accounts expand the risk if credentials are compromised.

What to check:

• Use dedicated integration users with narrowly scoped permissions that match only what the integration needs.
• Configure authentication using OAuth 2.0 or Named Credentials to prevent hard‑coding secrets.
• Rotate tokens and credentials regularly and avoid storing secrets in custom fields or hard‑coded Apex.

Issue #6. Webhook Reliability (Stripe)

What happens:

Stripe uses webhooks to deliver asynchronous payment and subscription events (e.g., payment success, subscription updates) to connected systems. Webhooks are best‑effort notifications and may not always arrive reliably or in order.

Typical failure modes:

• Webhook deliveries fail or time out, so the connected system never receives critical events.
• Events may be delivered more than once or out of order, requiring handling logic that accounts for duplicates and variability in delivery sequences.

What to check:

• Verify the webhook endpoint configuration in Stripe (correct endpoint URL, HTTPS, and event types).
• Ensure your integration processes events idempotently so that repeated deliveries do not produce inconsistent records.
• Confirm that webhook retries are being logged and monitored, and address connectivity or handler errors that prevent successful responses.

Best Practices for Reliable Salesforce Payment Integrations

Designing Salesforce payment processing with QuickBooks or Stripe Salesforce integration requires a careful, methodical approach. Integrations are not just about moving data; they ensure client records, subscriptions, and invoices remain accurate and consistent across systems.

Define the source of truth: Decide which system owns each type of data. Salesforce for client details, Stripe for subscription history, QuickBooks for invoices. This prevents conflicts and accidental overwrites.

Validate and map fields: Make sure all required fields in external systems have corresponding Salesforce fields. Map products, pricing, and plans correctly to avoid errors during record creation.

Monitor sync processes: Use logs, alerts, and reports to catch failures early. Partial syncs or missed updates can be costly if left unchecked.

Handle updates carefully: Changes to client data should flow in a controlled, predictable way. Implement error handling and plan for retries on failed API calls.

Secure the integration: Use secure authentication, tokenized payment data, and least-privilege integration accounts. Ensure compliance with PCI, GDPR, and other relevant standards.

Prebuilt AppExchange apps such as Breadwinner can connect QuickBooks Salesforce or Stripe with minimal setup. For organizations with specific business rules or complex requirements, custom API solutions provide additional flexibility.

Following these practices supports automating Salesforce workflows that reduce manual effort, prevent errors, and keep Salesforce and the connected financial system aligned, giving teams confidence in the accuracy of client and payment data.

Conclusion: Reliable and Scalable Salesforce Payment Workflows

Designing Salesforce payment processing with QuickBooks or Stripe Salesforce integration requires a careful, methodical approach. Integrations are not just about moving data; they ensure client records, subscriptions, and invoices remain accurate and consistent across systems.

Payment workflows work best when they are thoughtfully structured, monitored, and maintained. From planning triggers and mapping fields to handling updates and addressing real-world challenges, a clear and consistent approach keeps data aligned and processes predictable. When workflows are designed and managed properly, organizations can reduce manual effort, prevent errors, and maintain confidence in their customer and payment data.

By addressing common challenges and following proven recommendations, organizations can automate billing processes confidently, reduce manual effort, and maintain accurate customer and payment data across Salesforce, Stripe, and QuickBooks. This approach ensures that payment workflows are not only functional but also scalable and resilient as business needs grow.

Designing Salesforce Payment Workflows with QuickBooks and Stripe (Part: 1)

Guest blog by: Antonina Kharchenko

Introduction: Understanding Salesforce Payment Workflows

Managing client records and financial data across Salesforce and external systems is complex. Client data exists in Salesforce, but to create subscriptions, invoices, or records in Stripe or QuickBooks, teams often rely on spreadsheets, CSV exports, or emails. When a client updates an address, email, or billing detail, these changes must be manually reflected in Stripe or QuickBooks, increasing the chance of errors and delays.

Most business applications remain isolated: only about 29% of systems are integrated, and 95% of IT leaders report challenges with data integration. These disconnected systems force teams to rely on manual data transfers like spreadsheets and emails when synchronizing client records between Salesforce, payment providers, and accounting platforms.

This article focuses on how to design reliable payment workflows in Salesforce when integrating with systems like Stripe or QuickBooks. We’ll explore how to structure objects, define workflow triggers, model integration fields, and automate billing-related processes so that customer records, subscriptions, and invoices stay aligned across systems. These design principles help reduce manual work, improve data consistency, and support scalable billing workflows directly from Salesforce.

Designing Salesforce Payment Workflows

Payment Process, image from Breadwinner

A payment workflow in Salesforce is more than building Stripe or QuickBooks to Salesforce integration. It’s about structuring objects, fields, and automation so that client data, subscriptions, and invoices are created and updated reliably. Proper workflow design reduces manual work, ensures data consistency, and makes integration predictable.

1. Define the Workflow Triggers

• Choose the initiating events: Typically, a workflow begins with a Salesforce object changing state. For example:
• A new Account or Contact is created → trigger client provisioning in Stripe or QuickBooks.
• An Opportunity ready for billing → trigger subscription creation or invoice generation.
• Determine conditional logic: Decide when the workflow should execute, such as only for specific Opportunity types, regions, or billing plans. This avoids unnecessary API calls.

2. Model Integration Fields in Salesforce

• External system IDs: Store QuickBooks or Stripe customer IDs on the Salesforce Account or Contact object. This allows reliable mapping for updates.
• Payment and subscription status: Include fields such as Subscription Status, Invoice Status, Last Payment Date, or Last Sync Timestamp. These fields make it possible to monitor workflow execution and detect failures.
• Custom fields for required external data: Map required fields from QuickBooks or Stripe that Salesforce does not track natively (e.g., tax codes, billing cycles, Stripe plan IDs).

3. Decide on Automation Tools

• Flow vs Apex:
• Flow is suitable for standard automation, like creating a subscription after an Opportunity closes, or updating a QuickBooks record when an Account changes.
• Apex may be needed for complex logic, bulk updates, or handling retries for failed API calls.
• Process orchestration: Combine Flows with scheduled or triggered updates for tasks such as:
• Periodic reconciliation of invoices.
• Updating subscription status daily from Stripe.
• Keeping Salesforce recurring payment records aligned with external payment and accounting systems

4. Map Data Flow Direction

• Single source of truth: Define which system owns each piece of data. For example, Salesforce may own client contact info, while Stripe owns subscription history.
• One-way vs two-way sync: Determine if updates flow only from Salesforce → external system, or if external changes also flow back. One-way sync is simpler and reduces conflicts, two-way sync adds visibility but requires careful conflict resolution.

5. Sequence and Error Handling

• Workflow sequence example:
• Opportunity meets billing criteria.
• Validate client data in Salesforce.
• Push the customer record to Stripe or QuickBooks.
• Create a subscription or invoice in an external system.
• Record external ID and status back in Salesforce.
• Log errors or retries if any step fails.
• Monitoring and alerts: Include fields or reports to track failed updates, partial syncs, or API errors. Set up notifications for admins so issues can be resolved quickly.

Existing Solutions to Solve the Problem in Salesforce

There are several approaches to integrating Salesforce with financial systems to manage client data, subscriptions, and Stripe payments Salesforce. Each approach comes with trade-offs in terms of effort, reliability, and maintainability.

Summary: Each integration of Salesforce payment solutions has strengths and limitations. Organizations should weigh ease of setup, level of automation, and long-term maintainability when selecting a solution, for example to integrate QuickBooks with Salesforce.

Step-by-Step: Managing Salesforce Payment Workflows

To demonstrate how Salesforce payment workflows can be implemented, we reviewed available solutions on the AppExchange. Building a custom integration or using middleware would require more development time and maintenance. Prebuilt solutions provide a quicker way to get started while still illustrating the core workflow concepts. 

Salesforce Stripe integrations on AppExchange

After evaluating options, we selected Breadwinner, which can connect Salesforce with QuickBooks or Stripe. For this guide, we’ll use the Salesforce Payments Stripe app as an example.

Payments Integration on AppExchange

The focus of this guide is on synchronizing client records, creating subscriptions, and handling updates reliably, ensuring Salesforce remains aligned with the external system.

Step 1: Install the App

Install the Breadwinner Payments Integration from AppExchange into your Salesforce org. Assign the required permissions to integration users and admins as per the app’s setup guide.

Get the app

Step 2: Connect the Stripe

Use the app’s authentication flow to link a Stripe account. This establishes a secure connection so Salesforce can push and receive data.

Connect with a Payment Processor

Step 3: Configure Object Mappings

Map Salesforce objects (Accounts, Contacts, Opportunities, or custom objects) to Stripe customers and subscriptions. Identify which fields from Salesforce will populate the corresponding Stripe fields, such as email, billing address, and subscription plan.

Example of Associated Salesforce Account

Step 4: Test Synchronization

Test the workflow with a small set of records. Verify that client records, subscriptions, and related data are created correctly in Stripe and that updates in Salesforce are reflected as expected.

Create an Invoice or Subscription from Salesforce

Step 5: Monitor and Refine

Use app-provided logs and Salesforce reports to monitor workflow execution. Check for errors, partial syncs, or missing data, and adjust field mappings, triggers, or validation rules as needed.

Key Takeaway: Using a prebuilt AppExchange app allows you to connect Salesforce to an external system, synchronize client records, and automate subscription or invoice creation, keeping both systems aligned without complex coding.

Agentforce and Payment Integrations

Payment integrations can also support automation through Agentforce. When billing and payment data from Stripe or QuickBooks is synchronized with Salesforce, Agentforce can use this data to trigger workflows, update account status, or initiate follow-up actions based on payment activity.

Breadwinner provides integration apps listed on AgentExchange, the marketplace for Agentforce extensions within Salesforce. These apps allow synchronized financial data to be used reliably in Agentforce workflows and automation.

Breadwinner on AgentExchange

In practice, this type of setup allows teams to manage billing workflows directly from Salesforce, while payment processing and financial records remain managed in external systems such as Stripe or QuickBooks.

Closing Thoughts: Building Reliable Salesforce Payment Workflows

Payment integrations are not only about connecting systems, they are about designing workflows that keep customer, subscription, and financial data consistent across platforms.

When Salesforce is used as the operational hub for sales and customer management, integrations with platforms like Stripe and QuickBooks make it possible to automate billing processes while maintaining a clear and reliable source of truth for different types of data.

The key principles discussed in this article help reduce manual work and improve data reliability across systems. Whether organizations choose prebuilt AppExchange applications, middleware platforms, or custom integrations, the most successful implementations start with well-designed workflows inside Salesforce.

In the next article, we’ll explore the operational side of Salesforce payment workflows in more detail, including common integration challenges, typical failure scenarios, and practical recommendations for building stable and secure payment integrations.

121.82% Growth in Half a Year: Inside AgentExchange’s Explosive App Ecosystem

Guest blog by: Dorian Sabitov

Just one week before Dreamforce 2025, Salesforce’s AgentExchange – the marketplace for Agentforce AI components – surpassed the milestone of 100 public listings. This accomplishment underscores the rapid expansion of Salesforce’s latest ecosystem initiative and the growing engagement of partners in shaping its direction.

AgentExchange serves as the counterpart to Salesforce’s long-standing AppExchange but is centered around agentic components. These consist of prompts, actions, topics, and complete agent templates that customers can directly import into Agent Builder and extend throughout their Salesforce environments.

Since its launch in March 2025, AgentExchange has rapidly positioned itself as a key hub for AI innovation within the Salesforce platform. Guided by Salesforce leadership, the marketplace has experienced steady growth, balancing customer demand with contributions from its early partners.

AgentExchange by the Numbers: October vs. March 2025

Marketplace Overview

At the time of launch, specifically during the AgentExchange snapshot done on March 4, 2025, the marketplace featured 55 apps developed by 50 unique providers. The top three business categories were Sales (18 apps), Productivity (10 apps), and Finance (7 apps), emphasizing the platform’s strong focus on revenue acceleration, workflow efficiency, and financial connectivity.

Another snapshot of the marketplace, AgentExchange snapshot done on October 8, 2025, revealed 122 unique apps from 102 unique developers, showing that the marketplace more than doubled in six months (121.82% growth in apps and 104.00% growth in developers). Key business categories also shifted: Sales led with 49 apps, Productivity followed with 22 apps, and Analytics moved into third place with 12 apps.

Leading Developers

As of October 2025, the top AgentExchange publishers are:

• Bullhorn – 4 listings
• Breadwinner – 4 listings
• Salesforce Labs – 4 listings

In comparison, back in March there were only two publishers with three listings (the highest number of apps at that time):

• Salesforce Labs – 3 listings
• OpenText Corporation – 3 listings

As Stony Grunow, Co-Founder of Breadwinner, stated:

“AgentExchange is still in its early days, but thanks to the vision and dedication of leaders like Trish Phillips and Amy Gorman, partners like Breadwinner have a platform to innovate and expand what’s possible for Salesforce customers. Our mutual success is tied together — the stronger the partner ecosystem, the stronger Salesforce becomes.” 

Growth Timeline of Apps and Developers on AgentExchange

Below is a month-by-month overview based on snapshots. It presents total counts, month-over-month changes, and the average number of apps per developer.

Key Takeaways for AgentExchange Users

The AgentExchange app catalog is now large enough to support meaningful pilots without becoming overwhelming. From March to September, listings climbed to 114, with the strongest growth in Sales, Analytics, and Productivity.

August showed a brief dip due to cleanup, followed by September offering the widest selection. Use this pattern to schedule trials strategically and keep shortlists focused.

How to Run an Effective 30-Day Pilot

1. Choose one outcome, not several. For instance, reduce manual call outcomes or generate weekly pipeline summaries.
2. Shortlist 5 listings in the relevant category, then narrow down to 2 finalists after a 15-minute demo each.
3. Define a single success metric before installation. Examples are provided below.
4. Use a sandbox or development org, load a small test dataset, and grant permissions only to the pilot group.
5. Enable just one write action initially; keep all others read-only.
6. Review progress weekly, then decide whether to scale, refine, or stop.

Best Time to Start

• June and July snapshots were stable, making them ideal for pilots that prefer fewer surprises.
• September offered the broadest selection after cleanup, suitable for side-by-side comparisons.
• Expect occasional re-tags or removals during ongoing curation. Schedule weekly reviews of active listings.

Data and Security Checks Before Installation

• Review the objects and fields the agent reads and writes, including any new fields it creates.
• Verify permission sets required to run the agent and access outputs such as summaries or tags.
• Ensure field-level security for sensitive data, including revenue or PII.
• Check logging and audit details: where actions are recorded and how to roll back if needed.
• Understand rate limits and governor limits when the agent runs scheduled actions.
• Confirm the de-installation process: how to cleanly disable prompts, flows, and packages.

Success Metrics That Clearly Demonstrate Value

• Sales: time to update call outcomes, email preparation time per meeting, and number of meeting notes created per week.
• Analytics: accuracy of weekly pipeline deltas, time to produce a forecast note, and number of actionable insights per report.
• Finance: invoice status sync accuracy, time to resolve payment inquiries, and exceptions caught before close.
• Service: first reply quality score, average handle time for common cases, and daily use of agent-generated summaries.
• Marketing: campaign brief generation time, approved copy rate, and consistency of CTAs across assets.

Key Questions to Ask Vendors

• Which records does the agent update, and how are conflicts handled when multiple users act simultaneously?
• Can prompts and actions be versioned and transferred between sandboxes and production without manual edits?
• What occurs during an API outage, and what retry logic is implemented?
• How many active customers are running this in Sales Cloud, Service Cloud, or Experience Cloud?
• What does a typical two-hour configuration involve, and which steps usually cause delays?

Implications for AgentExchange Partners

AgentExchange more than doubled in six months. Buyers focus first on Sales, Analytics, and Productivity, but there is still opportunity in emerging areas such as Commerce, Collaboration, and IT-Admin.

Strategic Focus Areas

• High Demand, Higher Competition - Sales: 47 listings, 41.2% share. Win by delivering one precise write action on a standard field, with a short setup and a clear demo prompt.
• Steady Growth, Good Room - Analytics: 12 apps, 10.5% share. Provide a weekly summary or delta note that writes to Opportunity or Account, including one verified accuracy check. Productivity: 18 apps, 15.8% share. Focus on list views, reports, and task updates that eliminate a daily step.
• Emerging Lanes - Commerce: 4 apps, Collaboration: 4 apps, IT-Admin: 1 app. Focus on one concrete action, such as creating a follow-up task, setting a status, or filling a picklist.
• Uneven Categories That Still Convert - Finance: 11 apps. Be explicit about status sync, exception handling, and retries. Customer Service: 4 apps. Target improvements in first reply or case disposition write-back.

Straightforward Product Ideas

• Sales: post-call note plus next step, writes to Task and Opportunity.
• Analytics: weekly pipeline change note, writes to a Notes record on Opportunity.
• Finance: invoice status check, reads status, and updates a single picklist on Account.
• Commerce: abandoned cart follow-up, creates a Task with ready text and fills a status field on Lead.

Key Takeaways

AgentExchange grew fast and maintained its structure:

• Started at 55 listings in March, crossed 100 a week before Dreamforce, and reached 122 in October.
• Publisher count climbed to 102.
• Sales led the pack, adding 29 listings to reach 47.
• Analytics tripled to 12 listings.
• Productivity rose to 18 listings and held steady.
• Commerce appeared later, reaching 4 listings.
• August saw a dip due to cleanup, then September rebounded.

AI isn’t the future – it’s already here. Teams that adopt AI save time, reduce costs, and accomplish more. AgentExchange provides a straightforward way to discover functional Agentforce components, integrate them into Salesforce, and achieve results without lengthy projects.

For tool selection, keep it simple: choose one outcome, run a two-week pilot in a sandbox, enable one write action, and track a single metric. Sales, Analytics, and Productivity now have sufficient depth to demonstrate value without extended setups.

For building for the catalog, stay focused: deliver a safe action, include a copy-paste prompt, provide a short setup video, define clear permissions, and include an uninstall note. This combination builds trust. With Breadwinner Integrations Inc., Salesforce Labs, and Bullhorn each holding four listings, there remains space for targeted releases to stand out.

3 Ways to Integrate NetSuite with Salesforce

3 Ways to Integrate NetSuite with Salesforce by Brian Newbold.

In our digital world, we can’t afford to spend time re-entering our Salesforce sales into a billable form in NetSuite. A swivel seat is excruciating both in time spent and accuracy errors. And when it’s time to automate, it seems like all suggestions start with “it’s an easy-to-use template” yet ultimately end up being custom-built solutions.

My company, like yours, is certainly a special snowflake – unique in how we process orders and report our sales and financial data. But aren’t snowflakes also nearly identical in their structure, form, and function? Surely when you look very closely you will see the many fine and unique details, but at the macro level, they’re pretty much just the same. Similarly, aren’t our business processes nearly alike at macro with a modicum of uniqueness? 

In the sense of integrating NetSuite with Salesforce, it seems our business needs across most companies fall into separate categories in about 3 different ways: 

• First and most basic, where your processes fit the sample templates and have low volume. Everything pretty much just works out-of-the-box with the lowest-cost tier middleware
• Second, where most of us fit, having a healthy volume of orders and reasonable depth of customizations. We’ve outgrown the templates and pushed the boundaries a bit 
• Third and most complex, the large-enterprise ERP with a few thousand products and hundreds of thousands if not millions of customers. High volume and highly customized, where you’ve hired a team of purpose-driven integration engineers.

So now with a general idea to which category you belong, which tier of software should you choose? Well, that can be a bit confusing depending on whose advice you take. A wise man once told me that if you ask a plumber to fix your house, you’ll get a plumbing solution. Ask a roofer to fix the same house and he’ll fix it with… well, you get the point. 

When you reach out to an SI consultant, you will certainly get a systems integrator solution; but is it a fresh solution or the same old templated rollout? Plumbing and roofing haven’t changed much since 2006, but what about software solutions? Keep in mind the iterative progress your proposed solution will make in the coming years and strive to find a balance of features and longevity, as any solution is likely to be your solution for some time to come.

Enough with the cautionary tales and down to the suggestions.

Tier 1: Basic functionality, no customization

In this tier, my preferred go-to solutions are Zapier, Celigo, or Mulesoft and making use of built-in templates. This approach will perform the wire-up of NetSuite and Salesforce but will be nothing but the basics. The cost will be minimized but functionality will remain far from ideal. Growth is a little tough too and is sure to look closely at the transaction-based pricing structure and make sure it will continue to fit your growth plans.

Tier 2: Robust functionality, good customization.

In this tier, I would consider Celigo or Boomi with initial SI stand up, or Breadwinner with SI optional. Celigo and Boomi being many-to-many iPaaS solutions are phenomenal when you need to wire up an entire BizApps ecosystem. However, they come with hidden costs associated with the necessity to use a great systems integrator (SI) and a longer timeline than most expect. The NetSuite and Salesforce orchestration is generally the most complex wire-up of anything you’re connecting and will require diligence.

A bit of a hidden gem in the Tier 2 space is actually not an iPaaS, but rather a purpose-built NetSuite and Salesforce orchestration suite called Breadwinner. My favorite part of Breadwinner is that they’re constantly adding features and enhancing their product. Since it’s hosted on the AppExchange, it’s hosted from Salesforce itself. 

The clear advantage with Breadwinner comes in when you realize you’ve got all the NetSuite data stored and reportable alongside Salesforce. It’s a product you deploy, rather than a project you embark on. Definitely worth a nod as they’ve saved me a ton of time and have given a far more complete integration in their free trial, rather than the many-to-many packages deliver in their end product.

Tier 3: Custom functionality, all customization.

In this tier, expect the best fits to be Boomi, Talend, and the like with dedicated staff and definitely an initial SI standup. This is where the heavy hitters come in that can handle millions of transactions and complex transforms. By far the costliest solutions but sometimes the only ones that can get the job done. These are full-fledged ETL tools and if you’ve got the resources, these are for you. 

Boomi in my experience has been the easiest to set up initially, but I’ve pushed it and hit the upper limits of transactions. Talend’s strengths are in delivering great integrations in distributed node environments. A bit more programmatic than drag-and-drop, but well worth it when you’ve got those unique environments. 

Where does your business fit in?

Salesforce Einstein – Where to start to experiment and understand Machine Learning?

by Jean-Michel Mougeolle, Salesforce MVP hall of fame, Salesforce Einstein Champion, SharinPix CEO.


What is the Einstein Champion program?
Let me start this blog with the Einstein Champions Program. The Einstein Champions Program is for Trailblazers that are passionate about the Einstein Platform and want to share their advanced knowledge with peers and evangelize the power of Einstein.

I have the chance to be part of those, certainly, due to the various have made around Einstein Vision at Dreamforce and in many Dreamin’ events. I’m convinced that Einstein Vision is a great way to start learning with Machine Learning in Salesforce.


Why starting by Einstein Vision?
First, it will make you understand very easily the benefits and the approach required by Machine Learning.

Second, you can easily play with it, FOR FREE!
For Free? You mean you don’t need any licenses?
No, you just have to install the Einstein Vision and Language Model Builder by Salesforce Labs, to start playing with it. The creation of models is free, and to test them you have up to 2000 predictions per month for free as well.

So where should we start to create our first model?
I will go with Einstein Vision Image Classification. It only takes a zip file with few images organized by labels in folders to start with something. Of course, you may have to gather enough images per label to get something working, and take care of image format, size and resolution. But if you plan only to create a model for testing, extracting images from some google search should be sufficient to have nice results.


Can you explain the basis of Image Classification?
Yes, for sure, Image Classification makes prediction to identify a picture from examples on which it has been trained. As an example, the model can recognize a cat from a dog if it has been well trained with enough dogs and cats pictures.


For our demo jam with SharinPix we have used images from google to create models to classify food pictures. The model can recognize hot-dog, pizza, burger, drinks, dessert, BBQ meat and more. That’s a good example on how to classify from image line of a menu to make them sorted automatically.

You mean that you can train a model that easily?
Yes, you just have to catalog enough images per label (100), construct a zip file with those and create a dataset with it. The UI from the Salesforce Lab package allows you to easily create a dataset from a zip file. Once you have a dataset, you can train a model from the same package. The model is the « engine » to create predictions.
Once you have a model, you can present a picture and the model will make prediction.


What can we expect to learn from that?
The limits of a poor dataset.  As an example, if you upload only white cats and only black dogs in a dataset, you will get a bad quality dataset. If you present then a black cat to it, it will certainly predict it as a dog.

Getting a good dataset is key, and it’s really easy to understand from example that is not working. As Image Classification is very visual, you can learn easily about the right and wrong approach around Machine learning.


What about Object Detection?
It’s quite the same principle than Image Classification, but it can detect many objects in a picture and get back with the position, the numbers and of course the probability associated to each recognition. The main usage for this is to automate retail execution from Shelf Display pictures.

Is that as easy as for the Image Classification?
Yes and no.
It doesn’t require different technology and it’s the same approach: create a dataset with pictures and train a model to get prediction. But if you need to label the pictures with bounding boxes representing all the objects you want to recognize.

So, in the example of retail execution, you may have to make it learn from shell display images where you have to draw boxes around each object you want to recognize, with the name of it. And this time you don’t need 100 pictures per label, but 200 bounding boxes per label across all the pictures used in the dataset. And the drawing of the box requires to be precise for a good prediction.


What are the main problems that can make you have a bad dataset?
The first is the bad quality of labeling. AI is basing is logic on the examples you feed it with. If you give him wrong examples, it will result in bad predictions. When you label hundreds of images, it’s easy to make mistakes. There, QA is mandatory to avoid any errors in the labeling.
The second, the diversity, frequency, and quality of images are key. You should not use images too angled or with too much light. And you may need as well to get the same frequency for each object to recognize across all the images in the dataset.


You seem to be very well experienced around that, does it come from what you have done with SharinPix?
Yes, we have provided the services to create tons of models for various big retail customers, but also from the company in other industries. We have labeled datasets that can recognize multiple hundreds of objects and with multiple thousands of images.

The quality approach is key in that kind of project, getting organized, having the right level of QA and a good understanding of the risk for each problematic met is really important.
We have constructed an app to help the team that wants to be serious about model making, model optimization, and model maintenance. We use it internally and provide the services around worldwide too many different companies.


Is that available on the AppExchange?
Yes, it’s part of the SharinPix App, but you can reach me for any question about Machine Learning and the app whenever you need!


So, can you recap the best thing to start with if you want to learn about Einstein?
Yes, the first one is if course trailhead, there is an incredible TrailMix that will make you learn a lot: https://sfdc.co/einsteinchampionstrailmix 

Then you can install the Model Builder provided by Salesforce Labs from the AppExchange:
https://appexchange.salesforce.com/appxListingDetail?listingId=a0N3A00000Ed1V8UAJ

And of course, if you want some help and get serious about Image Recognition you can rely on SharinPix App and Labelling Services: http://bit.ly/SharinPixAppExchange