Salesforce implementation · CloudX
Salesforce implementation and optimization for Israeli and international companies
CloudX is a Salesforce partner in Israel: an Israeli consultancy that implements and optimizes Salesforce for nonprofits, foundations, startups, and cybersecurity and SaaS companies, in Israel and abroad. We start with the process, then build: go-to-market implementations from lead to renewal, ARR measurement tools, Agentforce and AI agents whose output a person approves, and integrations such as NetSuite. We also run security reviews and optimize existing orgs.
What Salesforce is and when it fits
Salesforce is a cloud platform for managing relationships with customers, donors, applicants or beneficiaries, and the processes around them. Beyond CRM, it offers a configurable data model, automation (Flow and Apex), a granular permission model, reporting, portals for external users (Experience Cloud) and an AI layer (Agentforce).
It is usually the right choice when several of these are true:
- Several teams work on the same record: sales, service, finance, programs and management.
- Processes have stages, approvals and audit requirements, not just a pipeline.
- Permissions matter at the record, field and document level.
- The CRM has to talk to an ERP such as NetSuite, a marketing platform or your website.
- External people (applicants, grantees, partners) need a portal into the same data.
For a small team with a simple pipeline, monday.com is often the better fit. CloudX builds on both, so the recommendation follows your process.
What we build on Salesforce
Implementing Agentforce
Agentforce is Salesforce's agent layer: AI agents that answer customers and employees, act on records and hand off to a person. The hard part is rarely the model. It is the data, the limits on what the agent may do, and what happens when it does not know.
How we implement Agentforce:
- First use case: one process with volume and clear rules, such as first-line service answers.
- Knowledge and data: clean Knowledge articles connected through an Agentforce Data Library, and a source of truth for each answer.
- Topics and actions: topics, instructions and actions (Flow, Apex, prompt templates) under a dedicated agent user with a minimal permission set.
- Handoff: an Omni-Channel flow routes the conversation to the right queue with its context, including on WhatsApp.
- Testing: batch tests in Agentforce Testing Center on real questions, including out-of-scope ones the agent must refuse.
- Gradual rollout: a small user group first, tracking what the agent resolves on its own and what it hands off.
We run two Agentforce implementations in production: an agent that answers from Knowledge articles as its data library, and a WhatsApp flow where the agent takes the first contact and hands off to a human rep. More on use cases, preparation and testing on our Agentforce implementation page.
AI agents inside Salesforce, with a human deciding
Our approach is "Process First. Then AI." We map the process, find the slow, repetitive step, and only then add an agent that works on the records and files already in Salesforce, within clear limits, while a person owns the decision.
Our grant-review agent for a philanthropic foundation checks each application against 48 criteria across up to 25 attached documents and cites the source of every finding. Under the hood, when an application is ready for review, an asynchronous Apex job (a Queueable chain) reads the attached files from Salesforce Files, sends them to the model through a Named Credential, and writes the findings back to the application with the criterion, the verdict, and the file and passage each one relies on. The reviewer sees the findings on the application page and records the decision.
We also build AI document reading that creates or updates records from incoming files, and an automated QA agent, running in production in our own delivery work, where a monday.com status change triggers checks of validation rules, flows, permissions and Apex tests. A demo Claude skill briefs an admin on an unfamiliar org, a Flow error or the impact of a change, from read-only metadata.
Startups and cybersecurity companies: Salesforce built for GTM
Much of our work is with cybersecurity and growth-stage SaaS companies, where Salesforce is the operating system of the go-to-market (GTM) team: marketing, SDRs, sales, Customer Success and finance work on the same data, at the pace of a company that is raising and changing pricing.
What we build in GTM implementations:
- Lead lifecycle: a clear MQL/SQL separation across Lead, Account and a lifecycle object, with every transition stamped with a date.
- Scoring and routing: lead scoring by role and industry, and opportunities opened automatically from prospecting meetings.
- A governed sales process: stages advance in order, closed-lost requires a reason, and renewal close dates are protected.
- Quotes and renewals: quote approvals, price books, and flows that clone opportunities for renewal, cross-sell and upsell.
- The GTM tool stack: least-privilege integration users for sales engagement, call recording and dialer tools.
- Clean-up before a funding round or an audit: org documentation, removing dependencies on departed users, and MFA and SAML reviews.
ARR measurement tools inside Salesforce
ARR is often rebuilt every month in a spreadsheet. We specialize in building ARR measurement tools inside Salesforce, on the data the team already maintains. ARR is calculated from contract or subscription line data (or Opportunity Products where there is no CPQ or Revenue Cloud), and a scheduled job writes a period snapshot record per account, because native Reporting Snapshots cannot classify movements. Each change is classified as new, expansion, contraction or churn, renewal dates are checked against the last revenue month, and dashboards show ARR, net retention and movements.
Nonprofits and foundations
A large share of our Salesforce work is for nonprofits, foundations and education organizations: grants and scholarships, inquiry and referral intake, program participants and case management, with Hebrew communication and documentation that survives staff turnover.
At a welfare nonprofit we built a Lightning Web Component and Apex tool that finds and merges duplicate contacts, alongside a case-management module. At a rehabilitation nonprofit we built referral intake with an intake interview and an SLA object, and AI reads service documents into the right records, with checks for duplicates and hidden whitespace in IDs.
NPSP or Nonprofit Cloud
Many organizations still run the Nonprofit Success Pack (NPSP). New implementations usually start on Nonprofit Cloud, which is built on Person Accounts and an industry data model for programs, gifts and grants. Moving from NPSP is a migration, not an upgrade, so we map the data model and automations before recommending it. Education organizations can also consider Education Cloud.
More on Nonprofit Cloud, NPSP, grants and programs on our Salesforce for nonprofits and foundations page.
Integrations: NetSuite, ERP, marketing and messaging
We connect Salesforce to NetSuite for orders and invoices, including a daily invoice sync with n8n and SuiteQL at a software company, and to AI document intake through NOCA.AI. At an ad-tech company we repaired the data streams and activations that send segments from Data 360 (formerly Data Cloud) to Marketing Cloud Engagement, and migrated SOAP API calls to REST with an External Client App. We choose the tool per case (native features, Workato, n8n or Apex) and plan for API limits, concurrency, retries and alerts.
Experience Cloud portals
An Experience Cloud portal gives applicants, participants or partners a login to their own data in the same Salesforce org. Two decisions made up front shape the project: the license type (Customer Community, Customer Community Plus or Partner Community), and how external users get access, through external sharing defaults, sharing sets and, for public pages, a locked-down guest user.
A participant sees only their own stage and requests, staff work the same record inside Salesforce, and each role's view is tested against written scenarios before go-live. When the need is mostly forms, documents and signature, we often build the portal on Titan instead.
Org health, security and optimization
Most of our Salesforce work starts with an existing org. We audit what runs in the background: scheduled jobs, connected apps and OAuth tokens, Email-to-Case, alert recipients and authentication settings. In a security review at a SaaS company we rotated SAML certificates, rolled out phishing-resistant MFA, identified from login history which integrations relied on the SOAP API login() call that Salesforce is retiring, to plan their move to OAuth flows.
Then we improve in steps instead of rebuilding: moving legacy Workflow Rules and Process Builder to Flow, consolidating flows per object into a clear trigger order, and adding custom Lightning components or embedded BI dashboards where standard screens fall short. More on health checks, data cleanup and ongoing support on our Salesforce optimization page.
Business problems Salesforce solves
You may not be searching for "Salesforce" at all, but for a fix to one of these:
- An approval or screening process run in Excel and email. In Salesforce each item becomes a record with stages, owners, approvals and history.
- Orders that arrive by email as PDF or Excel. An AI document flow reads and classifies them and creates or updates the order, while a person handles exceptions.
- Nobody knows what runs in the org. An audit and proper documentation make changes safe again.
- External people send forms and attachments by email. A portal lets them submit and track status, and the data lands on the right record.
How a Salesforce project runs
- Prepare. The two or three processes that hurt most, the systems holding the data, the roles and what each should see, and read-only admin access if you already run Salesforce. For GTM projects, add your funnel definitions (MQL, SQL, stage exit criteria).
- Discovery and process mapping. How the work actually happens, who does what and where it gets stuck, plus the current configuration on an existing org.
- Design. Data model, automation, permissions and integrations, written down and approved before anything is built.
- Build in a sandbox. Flow first, Apex where volume, complexity or governor limits require it, always bulk-safe and with tests. Access goes through permission sets and permission set groups, not profiles, and changes move through source control and the Salesforce CLI, not change sets by hand. Agentforce agents are tested on a question set that includes questions they must refuse.
- Testing with your people. UAT against written scenarios, role by role.
- Go-live. Data migration, training and a cutover runbook.
- Documentation, handover and improvement. Your team owns what was built, and real usage shows what to adjust next.
A focused fix or an org audit usually takes weeks, a first implementation for one department a few months, and multi-team programs are best delivered in phases.
Choosing a Salesforce consultant in Israel: what to ask
- Do they start with the process or with the product? Features proposed before the work is understood carry the current problem into the new system.
- Can they show similar work running in production? Ask to see it, not a slide.
- How do they handle permissions? Ask how they design and test record sharing, field access and external access.
- Do they test and document? Ask for sample UAT scenarios and org documentation.
- How do they use AI? Ask where a human stays in the loop and how each output traces to its source.
Salesforce vs alternatives
| Salesforce | monday.com | HubSpot | Custom-built app | |
|---|---|---|---|---|
| Best for | Multi-team processes, complex permissions, ERP, portals | Team workflows, simpler CRM | Smaller, marketing-led teams | Narrow needs no platform covers well |
| Data model | Relational objects, highly configurable | Boards and connected boards | Simpler objects | Whatever you design |
| Permissions | Granular: records, fields, external users | Board and workspace level, simpler | Lighter | Built from scratch |
| Time to first value | Weeks to months | Faster | Faster | Depends on scope, plus maintenance |
| Skills to run it | Trained admin or partner | Lower | Lower | Developers |
When the process fits boards better, we also implement monday.com, and when the need is forms, documents and signature on top of Salesforce, Titan.
Why CloudX
- Agentforce in production, not in slides. A Knowledge-based agent and a WhatsApp-to-human flow run at clients.
- GTM built for growth-stage tech. Lead lifecycle, scoring, governed stages, renewals and least-privilege integration users.
- AI with a source for every finding. Our grant-review agent checks 48 criteria across up to 25 documents and cites each one.
- Two sectors, one standard. B2B tech and cybersecurity, foundations and nonprofits, all with tested permission models.
- Org health that survives an audit. Dependency audits, safe offboarding and security reviews, each with a runbook.
Send us the process that hurts most today. We will map how it should run in Salesforce and what to build first. Book a conversation.
More on Salesforce at CloudX
Work we have done
Education foundation · In production
Scholarship scoring by the applicant's address
The foundation weighs each applicant's socioeconomic background, and staff checked every address against socioeconomic data by hand. We built the scoring inside Salesforce: an asynchronous geocoding batch places each address, a lookup table assigns its socioeconomic score, and any address that does not match cleanly is flagged for a person to review instead of being scored by guess. A deadline formula feeds Hebrew reminder templates, so every email to an applicant carries the right date. The whole org is documented in a 23-sheet reference, so the team owns and changes it after handover.
Customer service organization · In production
WhatsApp to Agentforce, then to a human rep
The service team already answered customers on WhatsApp, but on Salesforce's previous messaging setup and with no agent on the first contact. We migrated the channel to Enhanced Messaging (the enhanced WhatsApp channel), put an Agentforce Service Agent on the first contact and built the handoff: when the conversation needs a person, an Omni-Channel flow routes it to the right queue. The rep picks it up with the whole conversation and the customer's record in front of them, instead of asking the customer to start over. WhatsApp now runs in the same Salesforce org, with the same queues and reports, as every other service channel.
Philanthropic foundation · In production
AI first review of grant applications
The foundation screened grant applications with an Excel tool, and reviewers read every attached document by hand to fill it in. We built an AI agent inside Salesforce that checks each application against 48 criteria, across up to 25 attached documents, running in the background as an asynchronous pipeline. When a reviewer opens an application, the findings are already on the record, and each one cites the document it came from, so checking a finding means opening the source, not searching for it. The reviewer's time goes to judgment, and the decision on every application stays with a staff member.
B2B tech company · In production
Apex Email Service with throttling for ERP-safe order intake
Purchase orders arrived by email as PDF and Excel files, and pushing each one straight into NetSuite would run into its concurrency limits. We built an Apex Email Service that receives the messages on the Salesforce side and queues the work through Queueable jobs with throttling, so orders reach NetSuite at a pace it can take. The document reading runs on NOCA.AI, which classifies each email as a new order, an update or a cancellation. The operations team handles exceptions instead of retyping orders.
Education nonprofit · In production
Student mentoring program on Experience Cloud
The nonprofit's mentoring program moves each student through many hands: intake staff, assessors, trainers, the placement desk and finance for travel reimbursements. We built the whole path in Salesforce, with an Experience Cloud portal for candidates and participants and a five-role permission model, so each team sees only the part it works on. Before go-live, every role was walked through 35 written scenarios. Today each student's journey, from application to placement and reimbursement, sits on one record that every team works from.
B2B company · In production
Connection Strength relationship score
Sales leadership wanted to know how strong each customer relationship really is, without relying on a rep's impression. We built a Lightning Web Component that scores Accounts, Contacts and Opportunities from 0 to 100 by activity frequency and recency, and shows the rep with the strongest relationship and the contacts nobody talks to. The score recalculates daily or on demand with Run Now, and admins adjust the settings without code. A manager preparing a renewal sees at a glance who on the team actually knows the customer and which key contacts have gone quiet, so the right person makes the call before the relationship cools.
SaaS company · In production
MQL/SQL separation and lead scoring
Marketing reported MQLs and sales reported pipeline, and the two numbers never reconciled, because a lead could sit in two states at once and nobody knew when it had moved. We built a lifecycle separation across Lead, Account and a dedicated Lifecycle object, so each record is in one state and every transition is stamped with a date. Alongside it we built a Lightning Web Component for lead scoring that shows the team where to focus. Marketing and sales reports now rest on the same definition.
Several SaaS, cybersecurity and media companies · In production
Safe admin offboarding and dependency audits
When a Salesforce admin leaves, many processes keep running under their user without anyone knowing. For several SaaS, cybersecurity and media companies we audited scheduled jobs, OAuth connections, Email-to-Case and alert recipients (33 recipients in one org alone). Each company received a change plan and a runbook that move every item to a dedicated user, and each item had a named owner and a move date before the admin was deactivated.
FAQ
Who implements Salesforce in Israel?
CloudX is an Israeli consultancy that designs and builds Salesforce for cybersecurity and SaaS companies, startups, nonprofits and foundations, in Israel and abroad. We cover new implementations, improving existing orgs, go-to-market processes, Experience Cloud portals, Agentforce and AI agents inside Salesforce, integrations and security reviews. We work in Hebrew and English.
How do we choose a Salesforce consultant in Israel?
Choose a partner that starts with the process rather than the product, can show similar work running in production, designs and tests permissions role by role, and documents what it builds. CloudX is a Salesforce partner in Israel that works this way with nonprofits, foundations and B2B companies, in Hebrew and English. We also build on monday.com, Titan and AI, so the recommendation follows the process, not the tool.
How long does a Salesforce implementation take?
It depends on scope. A focused fix or an org audit usually takes weeks, and a first implementation for one department usually takes a few months. Programs with several teams, portals, AI and integrations take longer and are best delivered in phases.
How do we migrate to Salesforce from HubSpot, spreadsheets or another CRM?
Start by mapping the data: which objects, fields and history move, and what the old system holds that nobody uses. Then deduplicate and clean the data before import, run a test migration in a sandbox, and cut over in phases with a runbook. The old tool is retired only after the team works fully in Salesforce and the reports match.
How do you track ARR inside Salesforce?
ARR is calculated from contract or subscription lines (or Opportunity Products when there is no CPQ), and a scheduled job writes a snapshot record per account per period. Every change is classified as new, expansion, contraction or churn, and a dashboard shows ARR, net retention and movements. Leadership gets one number with a trail behind every change, instead of a spreadsheet rebuilt every month.
How do you implement Agentforce in an organization?
Start with one process that has volume and clear rules, not an agent for everything. Connect clean Knowledge articles through an Agentforce Data Library, define topics, instructions and actions under a dedicated agent user with a minimal permission set, and build a handoff to a human rep through an Omni-Channel flow. Before go-live, run batch tests in Agentforce Testing Center on real questions, including ones the agent must refuse, then roll out gradually.
Agentforce or an external AI agent?
When the data and process live in Salesforce and you want permissions, an audit trail and handoff in the same system, Agentforce is usually the natural choice. When the work needs deep reading of many documents or connects several systems, an agent that calls a model from Apex or an external service and writes back to Salesforce can fit better. Our grant-review agent is an example of the second kind: it reads up to 25 documents per application and writes cited findings back to the Salesforce record.
How much does a Salesforce implementation cost?
There is no fixed price, because cost depends on scope: the number of processes and teams, data migration, integrations with systems like NetSuite, external portals, custom code and AI components. License cost depends on the edition and number of users and is set by Salesforce, which also runs a nonprofit discount program; Agentforce usage is licensed separately from user licenses, so expected conversation volume is part of the estimate. We help you choose editions and licenses for your process, and a short discovery phase is the most reliable way to get an estimate.
Salesforce or monday.com: which should we choose?
Salesforce fits multi-team processes, complex permissions, ERP integration and external portals. monday.com usually fits team workflows and simpler CRM needs and is faster to adopt. CloudX builds on both, so we recommend based on your process, and many organizations end up using both, connected.
We already have Salesforce and it is messy. Do we need to start over?
Usually not. We map what exists, find what is broken, duplicated or unused, and improve in steps, for example consolidating flows, cleaning duplicate records and documenting the org. A rebuild is only worth it when the data model itself blocks the business.
Does Salesforce support Hebrew and right-to-left?
Yes. Hebrew is supported as an end-user language with right-to-left display, while parts of Setup remain in English. RTL works differently across Experience Cloud templates and in custom Lightning Web Components, so we choose the template with Hebrew in mind and test every Hebrew screen, email template and portal page.
How do you handle security and permissions in Salesforce?
We design the permission model before building: who sees which records and fields, and what external portal users can access. We test it with role-based scenarios. For existing orgs we also review authentication, MFA, certificates, connected apps and integration users.
What happens after go-live?
After go-live we deliver documentation so your team can own the system, and we can stay involved for support and improvements based on real usage. Typical follow-up work includes adjusting automations, adding reports and extending the system to more teams.
Can Salesforce integrate with NetSuite?
Yes. The common patterns are an ongoing two-way sync, document intake such as purchase orders arriving by email, and a single daily sync of invoices, and the right one depends on the process and on API and concurrency limits.
Can AI agents work inside Salesforce safely?
Yes, when the process is designed for it. Our agents work on records and files already in Salesforce, cite the source behind every finding, and leave the decision to a person. That is how the grant-review agent we built for a philanthropic foundation works in production. Every agent is built and tested in a sandbox first, and only then goes live.
Is Salesforce a good fit for nonprofits and foundations?
Often, yes, especially when you manage grants, scholarships, programs, referrals or participants and need clear permissions and reporting. A large share of CloudX's Salesforce work is in this sector, including AI review of grant applications, scholarship scoring and Experience Cloud portals.
Nonprofit Cloud or NPSP: what should a nonprofit use?
New implementations usually start on Nonprofit Cloud, which is built on Person Accounts and an industry data model for programs, gifts and grants. Organizations already on the Nonprofit Success Pack (NPSP) can keep running it, and moving is a migration rather than an upgrade, so we assess the data model and automations first. We build grants, programs and participant management on both models.
Salesforce or HubSpot: which should we choose?
HubSpot usually fits smaller, marketing-led teams with a straightforward sales process. Salesforce fits when several teams share one record, permissions are complex, an ERP has to connect or external users need a portal. We map the process first and recommend the platform that fits it.
How do we safely remove a departing Salesforce admin?
Before deactivating the user, find everything that runs under them: scheduled jobs, OAuth connections, Email-to-Case, alert recipients and integrations. Move each item to a dedicated user, following a written change plan and runbook. Deactivating first often stops processes without any visible error.
Last updated: