Ask most Sales Engineering leaders which deals are technically at risk this quarter, and the honest answer is that they find out in the forecast call. Not before it. This post explains what MCP for PreSales teams actually means, what the Elvance MCP Connector for Claude does about that blind spot, and what to check before you switch it on.

The information almost always exists. It is in a scorecard somebody filled in three weeks ago, a risk logged after a security review, a feature gap raised on a call that never reached the roadmap. It is scattered across records nobody opens between QBRs, and the person who could join it up is the same person running four evaluations this week.

That is the gap the Elvance MCP Connector for Claude was built to close. It puts your PreSales workflow inside a conversation, so the question “what is at risk?” gets an answer from Salesforce in seconds instead of a Friday spent chasing people.

The Admin Tax Nobody Budgets For

Every PreSales tool asks the same thing of the same busy people: open the record, find the field, type the update. It is not hard work. It is simply work that competes with the customer-facing work you hired them for.

The numbers show how little room there is. Consensus surveyed 423 PreSales professionals for its 2026 Sales Engineering Compensation and Workload Report and found that sales engineers spend 56% of their time on direct sales work. The other 44% goes to cross-functional work such as implementation support, marketing and product collaboration. When asked what would most improve their working life, the top answer was fewer repetitive tasks.

So the admin slips. The data goes stale, and the reporting built on top of it quietly stops being true.

Most vendors answer this with another interface: a separate workspace to keep updated alongside the CRM. That is a second system, and a second place for the truth to drift. Salesforce’s State of Sales 2026 research points the same way, with 51% of sales leaders using AI saying disconnected systems delay or limit their AI initiatives.

What Is MCP, and Why Does It Matter for PreSales?

MCP, the Model Context Protocol, is an open standard for connecting AI assistants to the systems where work actually happens. The MCP documentation compares it to a USB-C port for AI applications: one standard plug, instead of a custom cable for every tool.

Anthropic introduced MCP in late 2024. In December 2025, it donated the protocol to the Agentic AI Foundation, a fund under the Linux Foundation co-founded by Anthropic, Block and OpenAI, with support from Google, Microsoft, AWS, Cloudflare and Bloomberg. For buyers, that matters. You are not betting on one vendor’s proprietary plug-in format.

For PreSales teams, the practical change MCP brings is the direction of travel. Without MCP, you copy deal notes into a chat window and paste the answer back into Salesforce. With MCP, the assistant talks to the source directly, under your own permissions, and can act on your behalf. The record stays correct because the conversation is writing to the real thing.

Why AI Needs Context, Not Another Interface

Gartner predicts that by 2028 AI agents will outnumber sellers ten to one, yet fewer than 40% of sellers will say agents have improved their productivity. One of its recommendations to sales leaders is to build a central context layer that connects enterprise data, systems and seller judgment. Without it, agents produce generic answers instead of ones specific to your business.

PreSales has its own version of this gap. The same Consensus report found that 88% of SEs see some productivity gain from AI, but nearly 73% describe it as only slight or moderate. We saw the same theme at DEMOFEST London 2026. Our view is that most AI tools sit beside the work rather than inside it. They summarise a call well, but someone still has to move the outcome into the scorecard.

This is why being Salesforce-native matters for MCP for PreSales. Elvance stores scorecards, risks, feature gaps and activity on the Salesforce Opportunity records your team already uses. So when Claude updates a scorecard through the connector, it has updated the record the forecast is built on. There is no separate PreSales database to sync, and nothing to reconcile later.

What a Week With the Connector Looks Like

The full list of what the connector can and cannot do is on the connector page. Here is what it changes in two ordinary weeks.

For the Sales Engineer

It is Tuesday afternoon. You have just finished a technical deep-dive where the customer’s security team raised a data residency concern, and their architect asked for an SSO option you do not support yet. Normally that becomes three admin tasks for Friday, if it happens at all.

With the connector, it is one message to Claude: log the call, add the security concern as a risk, and record the SSO request as a feature gap. Claude finds the opportunity, shows you what it plans to change and waits for your confirmation. If your org pushes feature gaps to Jira or Linear, the gap reaches the product backlog as well.

For the PreSales Leader

It is Thursday, the day before the forecast call. Instead of messaging six SEs for updates, you ask Claude which opportunities have not been updated recently, which deals look at risk, and where the team’s time went this month. You see across the team, within the same visibility you already have in Elvance.

You walk into the forecast call with questions about specific deals, not a request for everyone to “update their opportunities”. Neither scenario needs a new tool. Each one removes a step between knowing something and recording it.

Five Questions Your Security Team Should Ask About Any MCP Connector

Any connector that can write to Salesforce will reach your security reviewer, and it should. These five questions cover most of what they will ask. They apply to any MCP connector, not just ours, so use them as a checklist when you compare tools.

Whose identity does it act as? A connector should act as the signed-in user, so it can never exceed that person’s permissions. Be wary of shared service accounts. Elvance runs every action through the user’s own Salesforce session.
What data can it reach? Ask for the scope in writing. The Elvance connector reaches Elvance data and fields only, never the rest of your Salesforce.
What can it change, and what stops a bad write? Look for validation, versioning and limits on deletion. Elvance rejects values outside your configured picklists, saves every scorecard change as a new version and cannot delete records.
What is stored, and where does it run? Ask what sits at rest and which sub-processors are involved. The Elvance connector keeps only short-lived sign-in tokens, runs on Cloudflare, and uses Claude as the AI, as listed in our privacy policy.
Who controls access? It should be off by default and easy to revoke. An Elvance admin must enable MCP Access and assign a permission set before anyone can connect.

A useful benchmark for remote connectors is the MCP authorization specification. It requires OAuth 2.1 security practices and explicitly forbids passing access tokens straight through to other services. Ask any vendor how they follow it.

Getting Started

Set-up takes two admin steps and one user step. Your Salesforce admin enables MCP Access and assigns the permission set, then each user adds the connector in Claude and signs in to Salesforce. You need an Elvance licence and a Claude plan that supports custom connectors. Full instructions and troubleshooting are on the Elvance Connector for Claude page.

Our suggestion: the best way to test MCP for PreSales is to start small. Pick one PreSales leader and two or three SEs, choose one workflow (activity logging, or the pre-forecast review), and run it for a month. That gives you a before-and-after view of how current your Opportunity data is before you roll it out wider.

Frequently Asked Questions

Is MCP Only for Claude?

No. MCP is an open standard, and the Elvance connector works with any MCP-compatible client. Claude is the primary supported client today, and the set-up instructions are written for it.

Can AI Write Bad Data Into Salesforce?

The connector rejects values that do not match your configured picklists, keeps a version history of every scorecard change, and cannot delete scorecards, risks or opportunities. Claude also confirms with you before it changes a record.

Does PreSales Data Leave Salesforce?

Your PreSales data lives in your Salesforce org. When you use the connector, the data you ask about is processed by Claude to answer your question, and the connector itself stores only short-lived sign-in tokens. Our privacy policy lists every sub-processor involved.

The Point of All This

PreSales teams drive revenue but often work in a blind spot. Leaders cannot see which deals are at risk, what it costs to win, or where effort is wasted, because the people holding that information are too busy to keep writing it down.

Removing the writing-it-down step is not a convenience feature. It is the difference between a forecast built on current data and one built on whatever was true a month ago. That is what MCP for PreSales should deliver, and it is the standard we hold our connector to.

Book a walkthrough and we will show you the connector running against real records.