Should we use Agentforce or build custom agents?

Three questions decide it. Where do the users already work? Where does the system of record live? Who must own the control plane? Agentforce when the work and the permissions live in Salesforce, including in Slack. Custom or Claude when the work spans other tools, or the data lives in ERP. Hybrid is common.

Do we need Data Cloud before Agentforce?

You need grounded, permissioned context. Sometimes that is Data Cloud. Sometimes it is clean Knowledge and a sharing model that already works. We will not quote an Agentforce build before that check.

Why do Agentforce and other AI pilots fail?

They die on context, permissions, and ops, not the model. Typical killers: quoting before a data-readiness pass, proposing more than about three agents in v1, shipping a demo org, and no transcript review after go-live.

Is every integration a MuleSoft project?

No. Point-to-point is honest when there is one pipe. Anypoint when you will reuse the asset. Flex Gateway when agents, MCP, or OAuth need a front door. We will say which.

Is Flex Gateway the same as Agent Fabric?

No. Flex Gateway is the edge. Agent Fabric is the control plane that inventories agents, MCP, and policies. You usually want both once more than one agent shares access.

Is IDP the same as an agent?

No. IDP structures a document. An agent may call IDP as a tool. A person still reviews money, identity, and policy holds.

Is Agentforce the same as Einstein Copilot?

No. Einstein recommends. Agentforce acts. If a partner conflates them, they have not shipped this.

Can you rescue a stalled AI pilot?

Both Salesforce orgs and Agentforce or custom-agent pilots that never left the sandbox. We assess, then finish, cut, or rebuild.

Do you publish a dollar menu?

No. A named scoped piece can be fixed-price after we have read it. An unread rescue cannot. Send the process.

Shivanath. The person who signs the work. There is no sales layer.

We already pay for Microsoft Teams. Why add Slack?

Keep Teams for meetings and files. Slack is for the work that crosses teams, partners and Salesforce: requests, approvals and handoffs with a record. Many companies run both, split by team.

Our team lives on WhatsApp. Will they use Slack?

Internal decisions move to Slack; customers stay on WhatsApp. We move one team and one process first, so people see the point before everyone switches.

Do our customers have to move off WhatsApp?

No. Customer conversations stay on WhatsApp. What moves is the internal chase behind them.

Can our Slack data stay in the UAE?

We confirm the data residency region for your plan with Salesforce before you sign, and write it into the proposal.

Do we need Salesforce to use Slack?

No. Slack works on its own, and some Slack plans include a built-in CRM. Most of our clients connect Salesforce because customers and deals already live there.

Agentforce or Claude in Slack?

Agentforce when the work and the permissions live in Salesforce. Claude when the job spans other tools. Both can work in the same channels, each on its own named job.

Is an AI agent in a shared channel safe?

It acts as a named member with limits you set, sees only the channels and data it needs, and a person approves anything that changes a record.

Can we invite brokers, agencies or partner banks into our channels?

Yes. Slack Connect shares a channel with other companies on your rules. We set who can invite outside firms and who approves a new one.

What happens to our messages when someone leaves?

The history stays with the company. Access is removed on the day, and retention rules decide how long messages are kept.

FAQ

What people ask before they buy.

No unpublished percentages. No invented headcount.

Answers

Straight answers. No percentages.

The assessment is free: a short read of the org and a written note. If you go ahead, the first build week is a fixed AED/SAR 14,900 (about USD 4,050), credited in full when you sign the build within 60 days.

Should we use Agentforce or build custom agents?: Three questions decide it. Where do the users already work? Where does the system of record live? Who must own the control plane? Agentforce when the work and the permissions live in Salesforce, including in Slack. Custom or Claude when the work spans other tools, or the data lives in ERP. Hybrid is common.
Do we need Data Cloud before Agentforce?: You need grounded, permissioned context. Sometimes that is Data Cloud. Sometimes it is clean Knowledge and a sharing model that already works. We will not quote an Agentforce build before that check.
Why do Agentforce and other AI pilots fail?: They die on context, permissions, and ops, not the model. Typical killers: quoting before a data-readiness pass, proposing more than about three agents in v1, shipping a demo org, and no transcript review after go-live.
Is every integration a MuleSoft project?: No. Point-to-point is honest when there is one pipe. Anypoint when you will reuse the asset. Flex Gateway when agents, MCP, or OAuth need a front door. We will say which.
Is Flex Gateway the same as Agent Fabric?: No. Flex Gateway is the edge. Agent Fabric is the control plane that inventories agents, MCP, and policies. You usually want both once more than one agent shares access.
Is IDP the same as an agent?: No. IDP structures a document. An agent may call IDP as a tool. A person still reviews money, identity, and policy holds.
Is Agentforce the same as Einstein Copilot?: No. Einstein recommends. Agentforce acts. If a partner conflates them, they have not shipped this.
Can you rescue a stalled AI pilot?: Both Salesforce orgs and Agentforce or custom-agent pilots that never left the sandbox. We assess, then finish, cut, or rebuild.
Do you publish a dollar menu?: No. A named scoped piece can be fixed-price after we have read it. An unread rescue cannot. Send the process.
Who do I talk to?: Shivanath. The person who signs the work. There is no sales layer.

The first decision

Agentforce or a custom agent.

Agentforce when

Custom or Claude when

The users already work in Salesforce, or in Slack connected to it.: The work spans other tools.
The system of record is Salesforce.: The data lives in an ERP or outside Salesforce.
Salesforce permissions should govern what the agent sees.: You must own the control plane yourself.

Hybrid is common. We pick from the process.

Slack

Slack, Teams and WhatsApp.

What companies ask before they move work into Slack.

We already pay for Microsoft Teams. Why add Slack?: Keep Teams for meetings and files. Slack is for the work that crosses teams, partners and Salesforce: requests, approvals and handoffs with a record. Many companies run both, split by team.
Our team lives on WhatsApp. Will they use Slack?: Internal decisions move to Slack; customers stay on WhatsApp. We move one team and one process first, so people see the point before everyone switches.
Do our customers have to move off WhatsApp?: No. Customer conversations stay on WhatsApp. What moves is the internal chase behind them.
Can our Slack data stay in the UAE?: We confirm the data residency region for your plan with Salesforce before you sign, and write it into the proposal.
Do we need Salesforce to use Slack?: No. Slack works on its own, and some Slack plans include a built-in CRM. Most of our clients connect Salesforce because customers and deals already live there.
Agentforce or Claude in Slack?: Agentforce when the work and the permissions live in Salesforce. Claude when the job spans other tools. Both can work in the same channels, each on its own named job.
Is an AI agent in a shared channel safe?: It acts as a named member with limits you set, sees only the channels and data it needs, and a person approves anything that changes a record.
Can we invite brokers, agencies or partner banks into our channels?: Yes. Slack Connect shares a channel with other companies on your rules. We set who can invite outside firms and who approves a new one.
What happens to our messages when someone leaves?: The history stays with the company. Access is removed on the day, and retention rules decide how long messages are kept.

Who asks these

Who brings us these questions, and what they hand us.

Example
Head of IT, Abu Dhabi insurance broker
Today: Asked whether Agentforce or a custom agent fits, and got two sales pitches.
After: Hands us the process and the systems. The answer comes from where users work and where the record lives.
Agentforce
Example
Operations manager, Dubai property management company
Today: Internal approvals run in WhatsApp groups, and tenants write on WhatsApp too.
After: Hands us one team and one process. Internal decisions move to Slack; tenants stay on WhatsApp.
Property management
Example
Integration lead, Jeddah manufacturer
Today: Every integration quote comes back as a MuleSoft project.
After: Hands us the list of pipes. Point-to-point when there is one; Anypoint when the asset will be reused.
Manufacturing

Examples of who we work with and why, not client stories.

After the questions

What happens after the answers.

01Week 1 in the orgWe read what already exists before anyone adds a cloud.
02Written noteWhat is worth fixing, what is worth finishing, and what is not. Not a dashboard.
03Fixed scope and priceWe can fix-price a named, scoped piece once the process is clear. We will not fix-price a rescue we have not assessed.
04BuildWe implement around the process you already run.
05We stayAfter go-live, a person stays on the exceptions.

The brief

Your question is not here?

The honest no: if the work is a software factory with a named delivery date, we are the wrong partner.