Go back to Market Research
Salesforce supplies customer relationship management software and applications for manufacturing, commerce, consumer goods, and field service. Its operational products maintain account forecasts, route customer orders, schedule technicians, manage promotions, and automate administrative processes. Agentforce Operations, derived from Regrello, adds configurable workflows spanning employees, suppliers, enterprise systems, and AI agents. 1 3 7 12 13 39
Overall supply chain score: 5.2/10
| Dimension | Score |
|---|---|
| Supply chain depth | 4.2/10 |
| Decision and optimization substance | 5.4/10 |
| Product and architecture integrity | 5.4/10 |
| Technical transparency | 6.4/10 |
| Vendor seriousness | 4.8/10 |
Salesforce combines substantial execution software with an incomplete supply chain decision model. Orders and schedules can run automatically, and some applications explicitly consider fulfillment costs or promotional margins. The governing logic nevertheless remains distributed across account forecasts, routing rules, scheduling policies, promotion models, and workflow blueprints. The documented products do not establish a common calculation of purchases, inventory deployment, prices, and service capacity under uncertain demand and supply. 1 5 6 10 13
Its agent strategy makes this distinction more consequential. Automating a supplier onboarding process or finding an order status can remove real work. It does not determine how much stock to buy or which demand should consume scarce inventory. Salesforce’s technical documentation often makes these boundaries clear; its promises of autonomous operations and effortless integration frequently obscure the engineering, configuration, and economic modeling still required. 14 15 16
Salesforce versus Lokad
Consider an online order that can ship from either a nearby store or a warehouse. Salesforce’s documented routing tools can reserve inventory, reduce shipment splits, and rank locations. Its September 2026 commerce release also advertises routing by fulfillment cost. Those computations organize the execution of an accepted order. 3 4 5
The harder economic question includes the next customer. Shipping the store’s last unit may save freight today while losing a higher-margin sale before the next replenishment arrives. Evaluating that choice requires a forecast of alternative future demand and receipt dates, together with the value of stockouts, transport, capital, and substitution. A routing policy needs those quantities to compare the alternatives on a common monetary basis.
Lokad’s inventory optimization offering builds recurring decisions around such economic trade-offs and probabilistic forecasts. Supply Chain Scientists implement and maintain the decision logic with the customer; that work belongs in its ownership cost. Salesforce supplies broader transactional applications and configurable automation, while the additional supply chain model must be established within the relevant product or connected to it. The distinction is where the economic reasoning runs and who improves it when decisions disappoint. Introduction to Supply Chain and Holimization explain Lokad’s design perspective.
What the operational products compute
Manufacturing forecasts organize commercial assumptions
Advanced Account Forecasting defines forecast contexts, dimensions, measures, and periods. Measures can come from user entries, formulas, or batch calculations. The Data Processing Engine extracts and transforms records, then writes values back into forecast objects. Templates support generating, recalculating, and rolling forecasts forward. This is a configurable calculation and collaboration framework; its data structures do not supply a probabilistic model of future demand or component availability. 1
Salesforce recommends Advanced Account Forecasting for new implementations and migration for existing Account Forecasting customers, with only one version enabled. The inputs can include opportunities, orders, sales agreements, historical orders, and customer-defined measures. A commercial commitment and a statistical forecast carry different uncertainty; combining them in a formula does not resolve that distinction. No joint demand-and-lead-time decision model is specified in these forecasting instructions. 2
Order routing has real actions and a limited economic perimeter
Distributed Order Management sends order summaries through automated flows, creates inventory reservations and fulfillment orders, and raises process exceptions when automation cannot complete. The standard routing example first finds the smallest sets of locations able to fill an order, then ranks those sets by average distance to its destination. These are executable routing decisions, with an explicit manual recovery path. 3 4
The September 2026 innovations page lists cost-based routing as generally available. It selects an eligible fulfillment location to minimize cost-to-serve while meeting the delivery promise. That is a narrower monetary objective than network-wide inventory optimization: the page supplies no formulation for the future opportunity cost of consuming stock, uncertain replenishment, or interaction with purchasing and promotions. It also declares manual fulfillment-rule maintenance finished, while describing merchants changing routing through natural-language prompts. Authoring changes faster leaves the responsibility for their correctness and economic effects intact. 5
Field Service contains a genuine scheduling engine
Salesforce’s 2025 scheduling paper describes a custom local-search metaheuristic. It starts from existing appointments, alternates diversification and improvement, and disrupts the incumbent schedule to explore alternatives. HERE provides mapping and traffic inputs. The same paper reports Pepsi Equipment Services conducting 10% more appointments in the same time with Enhanced Scheduling and Optimization, across an organization with over 2,500 technicians and approximately 500 dispatchers. This is specific operational evidence for scheduling, with no published net-return calculation. 7
The objective requires scrutiny. Customers weight service objectives such as travel, overtime, urgency, and resource preference. Enhanced Scheduling and Optimization minimizes penalty points; the older engine maximizes rewards. Salesforce’s example maps travel time onto a bounded scale before applying the weight. Such scores encode management preferences, but they do not automatically price the consequences of a delayed repair, an unsuccessful visit, or a scarce spare part. Different weights can select materially different schedules without showing which has the best expected economic return. 6
The implementation guide makes uncertainty largely an input-management problem. Customers estimate work durations, regularly review them, and set the earliest permissible appointment date to reflect parts availability. It also recommends narrowing scheduling horizons, eligible resources, or time-slot choices to reduce computation. Those restrictions can exclude valuable alternatives. The missing test is whether improved runtime preserves the value of flexible capacity when repair duration, new jobs, and part arrivals vary together. 8
Trade promotion models reach beyond dashboards
Consumer Goods supplies promotion budgets, margin comparisons, and templates with lift and cannibalization modeling. Its technical workflow clusters sales data, forecasts baseline volume, predicts promotional uplift, and writes forecast data back through Amazon S3. The training guide limits uplift predictions to 13 periods and 200 products and has the account manager request a prediction for a selected promotion. 12 10 11
These mechanisms address a substantive commercial problem. Their published workflow still does not show a joint choice of promotion, replenishment, and allocation under uncertain capacity and deliveries. A promotion can look profitable against its predicted uplift while exhausting the stock needed for other customers. Margin visibility and cannibalization estimates help describe that trade-off; the decision engine must actually retain it when selecting actions.
Agentforce Operations: process automation and its economic boundary
Agentforce Operations runs in a separately licensed workspace outside the Salesforce org. Blueprints specify stages, tasks, conditions, deadlines, and assignments to people or agents. Agents can extract information, compare documents, and validate data; users can approve work and supply missing information. Published blueprints can be versioned. This is a concrete process-execution architecture, with customer-authored logic at its center. 13
In September 2026, Salesforce described Marshall building processes without code or consultants and a Systems Integration Agent learning browser applications in about 90 minutes, then repairing screen changes in about 20 minutes. The same announcement reports automation of over 95% of Flex’s work-order approvals and roughly 70% of its third-party risk-assessment activities. It also proposes a multi-week proof of value against customer baselines for cycle time, automated work, recovered leakage, and hours saved. 14
That evaluation is more useful than a general promise of productivity. It must include the people who maintain rules, handle failed runs, investigate bad approvals, and update integrations. Approval automation measures the processing of an already defined task; it does not demonstrate that the approved work order is economically preferable to competing uses of materials and capacity.
Salesforce’s September 2026 enterprise-architect recruitment describes hands-on customer delivery: designing production integrations with ERP, PLM, and CRM, building agentic data pipelines, and adapting to evolving requirements. This is the material work omitted by the broad claim that consultants and traditional integration are unnecessary. A quick demonstration of one browser process cannot establish the lifetime cost of its production dependencies. 15
The MCP-based interface strategy adds another layer of access. Headless 360 exposes data operations, workflows, and existing services to agents; its August announcement distinguishes generally available servers from beta components. That can make reusable functions accessible, including batch computations. It does not supply the numerical economics absent from those functions. Reproducing a queue of human reviews through conversational calls remains a poor supply chain operating model even when the calls are automated. 36
Productivity, maintenance, and accountability
There is credible evidence of narrower labor gains. Salesforce attributes AAA’s five-minute average response-time reduction to its generative pre-work brief and applies that figure to six million annual roadside events. Axis Water reports 20% fewer return visits and technicians leaving for customer jobs 35 minutes earlier each day. These are useful measures of service execution. The resulting financial gain depends on staffing, additional work completed, and the cost of deploying and maintaining the system. 18
SharkNinja’s customer story describes automated order routing and fulfillment, alongside a wider commerce deployment involving Data 360, external data, and Salesforce support. It demonstrates real integration and operational use. Its commercial results cannot isolate the contribution of an inventory decision algorithm from storefront, merchandising, marketing, and service changes. 19
Salesforce’s February 2026 economic-impact summary is explicit about its construction: five interviewees informed a hypothetical composite retailer, with $200 million of annual online merchandise value, 119% modeled ROI, and a seven-month payback. The reported benefits include conversion and order-service efficiency. These are modeled economics for combined Commerce and Order Management, not observed supply chain returns for a named deployment. Applying that ROI to a replenishment project would substitute a different investment and benefit mechanism. 20
Configuration has its own engineering cycle. Field Service’s tuning guide asks customers to copy production-like data into a sandbox, establish baseline metrics, clone policies, vary weights, rerun simulations, and deploy first to selected territories. It supports controlled comparison and feedback. However, improving a weighted KPI within the existing scheduling problem is insufficient when an operating failure reveals a missing cost, uncertain repair outcome, or unavailable action. The model then needs reformulation and a fresh economic test, rather than another weight adjustment. 9
Agent usage introduces a further cost variable. The reviewed US list price is $500 per 100,000 Flex Credits; a standard Agentforce action consumes 20 credits, or $0.10 at that rate. The pricing page also offers conversation-based and user-based alternatives. These are alternative packages, not charges to add indiscriminately. The practical measure is the total cost of a completed, correct process, including retries, data services, implementation, and remaining labor. Counting more agent actions can mean more billing without more useful decisions. 38
Technology history: repeated revolutions and uneven continuity
The relevant history is broader than the current agent label. Salesforce has repeatedly presented new interfaces and infrastructure as transformations of enterprise work; delivery must be assessed product by product.
| Period | Promise and subsequent evidence |
|---|---|
| 2016: Einstein | Salesforce announced intelligence for every user and described Einstein as each customer’s data scientist. The named mechanisms included lead scoring, case classification, recommendations, and activity capture. Some were available; others were forthcoming. These bounded applications did not substantiate general enterprise intelligence. 29 |
| 2018–2020: Einstein Voice | The launch proposed spoken CRM updates, task creation, and intelligent record matching as a new productivity era. The Summer 2020 release notes scheduled retirement of the Voice Assistant beta for July 10, 2020. The initial interface revolution did not become a durable product in that form. 30 31 |
| 2019–2021: Blockchain | A Hyperledger Sawtooth-based low-code platform was announced for selected design partners, with general availability expected in 2020. The admin introduction, revised in May 2021, instead explains connecting blockchains using External Objects, MuleSoft, APIs, and partner applications. These sources do not supply the promised platform’s subsequent production scale or measured economic results. 32 33 |
| 2025: Agentforce migration | The original default agent stopped receiving new features and became unavailable in new environments from June 17, 2025. Salesforce directs customers to migrate to Employee Agent and warns that sequencing license changes incorrectly can interrupt availability. Agent adoption also creates migration work. 34 |
| 2025–2026: back-office acquisition and delivery | Regrello joined Salesforce in October 2025; the partner guide dates general availability of its Agentforce Operations products to February 6, 2026. The resulting workflow product has a defined operating model. Informatica, acquired in November 2025, expands data management; ownership does not itself unify the applications’ decision logic. 17 37 |
| 2026: MCP promotion and a narrower retreat | Headless 360 expands MCP access across products. Meanwhile, the Code Analyzer documentation says its MCP support ended in June 2026 and recommends CLI or editor tools unless an LLM is specifically needed, citing additional financial and resource costs. The retirement concerns Code Analyzer’s integration, not Salesforce MCP generally. Its reasoning is a useful corrective to the wider marketing narrative. 36 35 |
Salesforce produces working software across these generations. The credibility problem is the repeated leap from a bounded feature to a sweeping claim about how all enterprise work will change. A retired voice beta, unaccounted-for blockchain ambitions, and forced agent migrations make continuity and maintenance part of today’s proposition. New terminology cannot discharge the earlier promises.
Software materials and architectural cost
The following inventory separates infrastructure roles and product generations. It shows considerable engineering substance, alongside substantial operating mass.
| Component or subsystem | Documented role and scope |
|---|---|
| Salesforce forecasting objects and Data Processing Engine | Configurable forecast data and ETL calculations in Manufacturing. This is a different function from the Field Service optimizer. 1 |
| Field Service metaheuristics and HERE data | Scheduling search and location inputs described in the 2025 technical paper. 7 |
| CRM Analytics and Amazon S3 | Baseline/uplift model preparation and forecast writeback in the documented promotion workflow. 10 |
| Kubernetes, Terraform, Spinnaker, container images | Hyperforce’s December 2022 engineering account describes immutable deployment, infrastructure configuration, and rollout mechanisms. 25 |
| Trino on Kubernetes | Data Cloud’s July 2024 engineering account documents moving query workloads from EC2 to Kubernetes, including integration with Hyperforce’s modified cloud environment. 26 |
| Apache HBase/Phoenix and DynamoDB-compatible APIs | An October 2025 database-team account describes a common NoSQL service intended to reduce inconsistent SQL/NoSQL behavior and cloud-specific code across Salesforce infrastructure. 28 |
| Agentforce Operations backend and data integrations | The June 2026 engineering role describes tightly coupled domains being decomposed, complex GraphQL interfaces, and daily synchronization with Data Cloud, plus integrations involving MuleSoft, Flow, and Informatica. 16 |
The Operations hiring evidence is particularly useful: the integration boundary is still being engineered, and daily synchronization is an explicit workload. A workflow spanning that boundary must account for stale state and failed updates; a common brand does not make all observations current or all actions atomic. The posting’s preferred Go/PostgreSQL expertise is a hiring signal, rather than a verified inventory of every customer’s deployment. 16
Salesforce also quantifies an infrastructure problem itself. Its April 2026 Hyperforce article reports over 8,000 Kubernetes-based services and roughly 47% of managed Kubernetes spending tied to idle capacity. Configuration drift made corrections difficult. The team built an agent that prepares changes as pull requests, with separate deterministic optimization after initial LLM approaches struggled with fragmented configurations and simultaneous constraints. CPU-request changes pass through existing validation and staged deployment. 27
This is evidence both of expensive accumulated complexity and of a concrete repair. It does not establish a 47% waste rate for a particular customer’s subscription or the current platform after those changes. The broader lesson applies directly to supply chain agents: the language model can help locate and modify software, while a separate computation must select consistent decisions across interacting resources. Salesforce’s own infrastructure engineers articulate a more demanding standard than merely adding a conversational interface.
Security: enforced boundaries and conflicting assurances
Apex has inspectable controls that go beyond certifications. Database operations in API version 67 and later default to the current user’s object and field permissions; version 66 and earlier defaulted to system mode. This version distinction matters for existing custom code. Explicitly elevated operations and historical classes still need review when exposed to an agent. 21
Outbound Apex calls require a registered remote site or a named credential defining the endpoint. Runtime governors impose CPU and heap limits on transactions. These are real restrictions outside the submitted application logic, materially different from allowing arbitrary practitioner code unrestricted access to a production host. They do not make an excessively privileged business action safe. 40 41
Salesforce gives conflicting instructions about Agentforce data masking. Its June 2026 help article says Agentforce masks sensitive prompts before external model processing. The dedicated Agentforce limitations page explicitly disables pattern-based and field-based masking because it can impair response quality. It distinguishes external providers, covered by retention agreements, from models hosted inside Salesforce’s trust boundary. These arrangements do not provide the advertised removal of sensitive information before model processing. 22 23
Custom actions inherit the security behavior of their referenced Apex class, flow, or prompt template. Agentforce’s documentation also describes an allowlist for URLs and records prompts, responses, and trust signals in Data 360. The crucial boundary is the authority of each callable action: prompt instructions alone cannot contain the damage from an agent invoking a powerful action on the wrong records. Salesforce supplies controls, but configuring and maintaining those controls remains part of deployment. 24
Supply chain score
Supply chain depth
| Criterion | Score |
|---|---|
| Economic framing | 4/10 |
| Decision end-state | 6/10 |
| Conceptual sharpness on supply chain | 4/10 |
| Freedom from obsolete doctrinal centerpieces | 3/10 |
| Robustness against KPI theater | 4/10 |
Average: 4.2/10.
Economic framing — 4/10. Cost-based fulfillment and promotion margins provide local monetary reasoning. The missing future value of stock and shared resources prevents those local calculations from establishing system-wide economic optimization. 5 12
Decision end-state — 6/10. Order reservations, fulfillment instructions, and technician assignments reach execution. Account forecasts and selected-promotion predictions still leave major planning decisions to other software or people. 3 7 11
Conceptual sharpness on supply chain — 4/10. The products represent meaningful operational distinctions, including repair skills, commercial agreements, and promotional effects. Their common workflow infrastructure supplies no common treatment of opportunity cost across these decisions. 1 6 12
Freedom from obsolete doctrinal centerpieces — 3/10. Forecast collaboration, weighted scheduling KPIs, and task-by-task blueprints remain central. An agent executing that structure can save clerical effort while retaining the same fragmented planning responsibilities. 1 6 13
Robustness against KPI theater — 4/10. Several customer measures have intelligible operational meaning. The economic-impact study identifies its composite construction, but the headline ROI and automation rates cannot validate the profitability of purchases or allocations. 18 20
Decision and optimization substance
| Criterion | Score |
|---|---|
| Probabilistic modeling depth | 3/10 |
| Distinctive optimization or ML substance | 6/10 |
| Real-world constraint handling | 7/10 |
| Decision automation and labor productivity | 6/10 |
| Resilience under real operational complexity | 5/10 |
Average: 5.4/10.
Probabilistic modeling depth — 3/10. Expected durations, input forecasts, and promotional predictions do not establish joint uncertainty in demand, part availability, and execution. The resulting policies have no documented basis for pricing correlated shortages or disruptions. 2 8 10
Distinctive optimization or ML substance — 6/10. The scheduling search method and promotion prediction pipeline are substantive mechanisms. Their disclosed objectives and evaluation remain narrower than economically coupled supply chain decisions. 7 10
Real-world constraint handling — 7/10. Skills, work windows, territories, stock availability, and fulfillment splits shape executable choices. Shortening horizons or excluding candidates to ease search can sacrifice flexibility whose value emerges only under uncertainty. 4 8
Decision automation and labor productivity — 6/10. Customer examples document reduced service effort and working order automation. Total remaining customer, vendor, and partner effort is insufficiently measured to extrapolate those gains to an autonomous supply chain function. 18 19
Resilience under real operational complexity — 5/10. Sandbox comparisons and staged policy deployment provide a practical correction path. The documented loop mostly tunes an existing KPI formulation; economically harmful recommendations may require new objectives, uncertainty models, or actions beyond that loop. 9
Product and architecture integrity
| Criterion | Score |
|---|---|
| Architectural coherence | 5/10 |
| System-boundary clarity | 6/10 |
| Security seriousness | 6/10 |
| Software parsimony and ownership cost | 3/10 |
| Compatibility with programmatic and agent-assisted operations | 7/10 |
Average: 5.4/10.
Architectural coherence — 5/10. Shared platform services coexist with separate forecasting, analytics, routing, and workflow engines. The Operations backend’s documented hardening and integration work show why platform ownership cannot substitute for coherent state and decision semantics. 1 16
System-boundary clarity — 6/10. Manuals identify the separate Operations workspace, external fulfillment inputs, and integration mechanisms. Broad claims of seamless autonomy conceal more dependencies than these technical descriptions do. 3 13 16
Security seriousness — 6/10. Permission enforcement, outbound endpoint restrictions, and runtime governors provide meaningful containment. Older execution defaults, custom-action privileges, and the masking discrepancy remain material deployment risks. 21 23 24 40 41
Software parsimony and ownership cost — 3/10. Infrastructure sprawl, customer-specific integrations, policy maintenance, and metered agent work all consume resources. Salesforce documents ways to reduce those burdens, but the overall operating model remains heavy for the supply chain decisions demonstrated. 15 27 38
Compatibility with programmatic and agent-assisted operations — 7/10. APIs, callable workflows, controlled code, and headless services support substantive automation and software improvement. Their economic usefulness depends on the decisions implemented behind them. 13 36 40
Technical transparency
| Criterion | Score |
|---|---|
| Public technical documentation | 8/10 |
| Inspectability without vendor mediation | 7/10 |
| Portability and lock-in visibility | 4/10 |
| Implementation-method transparency | 7/10 |
| Evidence density behind technical claims | 6/10 |
Average: 6.4/10.
Public technical documentation — 8/10. Public manuals expose calculation methods, scheduling behavior, security boundaries, and migration requirements. They enable a much more concrete assessment than the product slogans, although the full numerical engines remain proprietary. 1 6 21
Inspectability without vendor mediation — 7/10. Tutorials and engineering publications let a reader inspect important assumptions and operating procedures. Published data and code sufficient to reproduce the principal customer economic claims are missing. 4 9 27
Portability and lock-in visibility — 4/10. External interfaces make data movement practical. Recreating forecast objects, policies, blueprints, permissions, and platform-specific behavior elsewhere remains a software migration, not a data export. 1 13 34
Implementation-method transparency — 7/10. The guides explain preparation, testing, policy changes, and integration responsibilities. The missing step is a reusable method for validating the complete economic model after operational failures. 8 9 15
Evidence density behind technical claims — 6/10. Concrete methods and candid engineering accounts support the operational capabilities. Claims about universal autonomy, effortless integration, and future inventory economics run ahead of the published demonstrations. 7 14 27
Vendor seriousness
| Criterion | Score |
|---|---|
| Technical seriousness of public communication | 4/10 |
| Technology roadmap integrity | 4/10 |
| Conceptual sharpness | 5/10 |
| Economic accountability | 5/10 |
| Defensibility in an agentic-software world | 6/10 |
Average: 4.8/10.
Technical seriousness of public communication — 4/10. Strong technical material sits beside overbroad integration claims and contradictory security documentation. Customers must resolve those inconsistencies before relying on the promised behavior. 14 15 22 23
Technology roadmap integrity — 4/10. Durable operational products coexist with discontinued assistants, required migrations, and poorly documented follow-through on blockchain. New agent releases carry a continuing support and migration obligation. 31 32 33 34
Conceptual sharpness — 5/10. Engineering accounts distinguish language-model assistance from constrained computation. The supply chain narrative less consistently distinguishes faster administrative processes from better economic decisions. 13 27
Economic accountability — 5/10. The Operations proof-of-value proposal establishes observable customer baselines, while the commerce study states its modeled scope. Public evidence still leaves net returns, bounded ongoing obligations, and responsibility for correcting poor supply chain decisions unresolved. 14 20
Defensibility in an agentic-software world — 6/10. Transactional applications, scheduling engines, execution controls, and deployed integrations remain useful when interfaces change. Their lasting value depends on reducing operating and maintenance work; exposing existing tasks through MCP alone does not establish that advantage. 7 13 35 36
Sources
[1] How Advanced Account Forecasting Works
How Advanced Account Forecasting Works. Publication date not specified. Accessed November 4, 2026.
[2] Forecast Your Business in Manufacturing Cloud
Forecast Your Business in Manufacturing Cloud. Publication date not specified. Accessed November 4, 2026.
[3] Distributed Order Management
Distributed Order Management. Publication date not specified. Accessed November 4, 2026.
[4] Implement Distributed Order Management
Implement Distributed Order Management. Publication date not specified. Accessed November 4, 2026.
[5] Commerce Cloud Innovations: Dreamforce 2026
Commerce Cloud Innovations: Dreamforce 2026. Published 2026-09. Accessed November 4, 2026.
[6] Create and Manage Field Service Objectives
Create and Manage Field Service Objectives. Publication date not specified. Accessed November 4, 2026.
[7] Field Service Scheduling and Optimization
Field Service Scheduling and Optimization. Published 2025. Accessed November 4, 2026.
[8] Guidelines for Effective Schedule Automation and Optimization
Guidelines for Effective Schedule Automation and Optimization. Publication date not specified. Accessed November 4, 2026.
[9] Fine-Tune the Scheduling Policy
Fine-Tune the Scheduling Policy. Publication date not specified. Accessed November 4, 2026.
[10] Set Up CRM Analytics to Predict Sales Uplift Volume
Set Up CRM Analytics to Predict Sales Uplift Volume. Publication date not specified. Accessed November 4, 2026.
[11] Set Up CRM Analytics to Predict Volume
Set Up CRM Analytics to Predict Volume. Publication date not specified. Accessed November 4, 2026.
[12] Trade Promotion Management
Trade Promotion Management. Publication date not specified. Accessed November 4, 2026.
[13] Optimize and Automate Business Processes with Agentforce Operations
Optimize and Automate Business Processes with Agentforce Operations. Publication date not specified. Accessed November 4, 2026.
[14] Your Business Processes Are Ready for Their Revolution
Your Business Processes Are Ready for Their Revolution. Published 2026-09-14. Accessed November 4, 2026.
[15] Senior Enterprise Architect: Agentforce Supply Chain
Senior Enterprise Architect: Agentforce Supply Chain. Published 2026-09-16. Accessed November 4, 2026.
[16] Agentforce Operations Engineering Job JR340462
Agentforce Operations Engineering Job JR340462. Published 2026-06-17. Accessed November 4, 2026.
[17] Agentforce Operations Partner Pocket Guide
Agentforce Operations Partner Pocket Guide. Publication date not specified. Accessed November 4, 2026.
[18] Field Service Transformed by Agentic AI
Field Service Transformed by Agentic AI. Published 2025-04-09. Accessed November 4, 2026.
[19] SharkNinja Unified Ecommerce
SharkNinja Unified Ecommerce. Published 2026-06-16. Accessed November 4, 2026.
[20] Forrester TEI: B2C Commerce and Order Management
Forrester TEI: B2C Commerce and Order Management. Published 2026-02-23. Accessed November 4, 2026.
[21] Enforce Object and Field Permissions
Enforce Object and Field Permissions. Publication date not specified. Accessed November 4, 2026.
[22] How to Mask Data in Salesforce
How to Mask Data in Salesforce — Overview of Data Masking Tools and Features. Published June 13, 2026. Accessed November 4, 2026.
[23] Data Masking Limitations in Agentforce
Data Masking Limitations in Agentforce. Publication date not specified. Accessed November 4, 2026.
Trust and Agentforce. Publication date not specified. Accessed November 4, 2026.
[25] Immutable Infrastructure in Hyperforce
Immutable Infrastructure in Hyperforce. Published 2022-12-14. Accessed November 4, 2026.
[26] Data Cloud EC2 to Kubernetes Migration
Data Cloud EC2 to Kubernetes Migration. Published 2024-07-22. Accessed November 4, 2026.
[27] Hyperforce Kubernetes Idle Capacity
Hyperforce Kubernetes Idle Capacity. Published 2026-04-30. Accessed November 4, 2026.
[28] Apache Phoenix DynamoDB Compatibility
Apache Phoenix DynamoDB Compatibility. Published 2025-10-07. Accessed November 4, 2026.
[29] Introducing Salesforce Einstein
Introducing Salesforce Einstein. Published 2016-09-19. Accessed November 4, 2026.
[30] Introducing Einstein Voice
Introducing Einstein Voice. Published 2018-09-20. Accessed November 4, 2026.
[31] Salesforce Summer 2020 Release Notes
Salesforce Summer 2020 Release Notes. Published 2020. Accessed November 4, 2026.
[32] Salesforce Blockchain Launch
Salesforce Blockchain Launch. Published 2019-05-29. Accessed November 4, 2026.
[33] Introduction to Salesforce Blockchain for Admins
Introduction to Salesforce Blockchain for Admins. Published 2021-05-12. Accessed November 4, 2026.
[34] Agentforce Default Considerations
Agentforce Default Considerations. Publication date not specified. Accessed November 4, 2026.
[35] Use MCP Tools to Analyze Your Code
Use MCP Tools to Analyze Your Code. Publication date not specified. Accessed November 4, 2026.
Expanding Headless 360. Published 2026-08-25. Accessed November 4, 2026.
[37] Informatica Acquisition Completed
Informatica Acquisition Completed. Published 2025-11-18. Accessed November 4, 2026.
Agentforce Pricing. Publication date not specified. Accessed November 4, 2026.
[39] Manufacturing Release Notes
Manufacturing Release Notes. Published 2026. Accessed November 4, 2026.
[40] Apex Remote Site Settings
Apex Remote Site Settings. Publication date not specified. Accessed November 4, 2026.
[41] Apex Execution Governors and Limits
Apex Execution Governors and Limits. Publication date not specified. Accessed November 4, 2026.