Skip to content
← All work

Salesforce AI Assistant

Account research and call prep, directly inside Salesforce.

Year
2026
Status
In development
Client
Superside
Metric
Internal AI workflow for reducing manual Salesforce account research and call prep.

Problem

Sales reps spend too much time preparing for calls in places that are not Salesforce.

A single opportunity might require checking the account record, contact history, company website, LinkedIn, past emails, call notes, CRM fields, sales methodology notes, and recent account activity. The information exists, but it is fragmented across too many tabs and too many systems.

That creates two problems. Reps lose time before calls, and managers get inconsistent account preparation across the team. The quality of prep depends on the rep, the available time, and how well the CRM has been maintained.

The opportunity was to bring AI into the workflow reps already use instead of asking them to visit another tool.

Approach

I built a Chrome extension that works directly on Salesforce opportunity, account, and contact pages.

The assistant scans the current Salesforce page, extracts the relevant context, and helps the rep prepare for the next action. It can summarize the account, identify useful research angles, surface likely stakeholders, prepare call notes, and suggest what the rep should understand before engaging.

The key design decision was to make the tool contextual. The user should not have to explain which account they are working on or copy-paste CRM information into a chatbot. The assistant should understand the page they are already on and generate useful support from that context.

This keeps the AI close to the work and makes adoption easier. It behaves less like a separate product and more like an extra layer of intelligence on top of Salesforce.

Outcome

The Salesforce AI Assistant is currently being built for internal use at Superside.

The expected impact is a faster and more consistent account-prep workflow for sales reps, especially when working through complex opportunities, unfamiliar accounts, or multi-stakeholder deals.

The broader thesis is that AI will be most useful in sales when it does not ask reps to change behavior dramatically. The best version sits inside the existing workflow, reads the context, and helps the rep take the next best action.

Stack

  • Chrome Extension - in-browser workflow layer
  • Salesforce page context - opportunity, account, and contact data
  • React / JavaScript - extension interface
  • OpenAI / Claude - research, summarization, and prep generation
  • Prompt orchestration - structured account research and call prep workflows
  • CRM data layer - Salesforce records and field context
  • Web research layer - company, contact, and market context