At a glance
- Industry
- B2B Marketing
- Platforms
- Anthropic Claude, Claude + CRM Integration, Salesforce Sales Cloud
- Service
- Integration
- Results
- May–Jul 2026 phased production rollout
The challenge
Abstrakt Marketing Group wanted to make sales-call information useful across appointment preparation, quality review, and coaching. The goal was to bring structured summaries and evaluations into the places where clients, account managers, and sales leaders already worked. The company also wanted practical applications for employee training and internal operations. The engagement connected those use cases to existing systems and company-specific processes rather than treating AI as a standalone tool.
What we did
- Call summaries and appointment preparation: built Abstrakt Intelligence using Claude on Amazon Bedrock to turn call transcripts into structured summaries and key highlights, surfaced in client appointment briefing emails and the account-manager appointment queue.
- Talk-track analysis: added reporting that compares calls with account-specific talk tracks, including introductions, recaps, closing, and objection handling, identifying areas for coaching based on how conversations align with the approved approach.
- Salesforce scorecards: built call-quality scoring against Abstrakt Marketing Group’s evaluation criteria, with grades, supporting transcript excerpts, and coaching notes written back to Salesforce, calibrated against human-scored calls.
- Interactive sales training: deployed Engage, an application that uses Claude for practice, feedback, and coaching aligned with the sales process.
- Internal workflow support: deployed a time-off management application with Claude-assisted request handling, plus custom MCP connectors that give staff controlled access to internal systems from Claude.
The results
Abstrakt Marketing Group put these capabilities into production in phases from May through July 2026. Call summaries and appointment highlights were live in June, followed by talk-track reporting and Salesforce AI scorecards in July. The deployment made structured call insights available in appointment preparation and put AI-generated quality evaluations into Salesforce, while sales representatives gained interactive practice and feedback through Engage and internal teams received additional workflow tools and controlled connections to business systems. These were deployed workflows, not demonstrations: a set of Claude-powered capabilities embedded in recurring sales and operational work.
Does this sound like your team?
- Account managers walk into client appointments without a quick summary of what was said on the call that set it up.
- Call reviews depend on a manager listening to recordings, so only a small sample gets scored and feedback arrives late.
- Reps rarely get to rehearse objections before live calls, and coaching notes sit in documents nobody opens from Salesforce.
Where we would start
Begin with one use of call transcripts that already has a home, such as the briefing someone reads before an appointment. Next, write your scoring criteria down explicitly and compare AI grades with calls your managers scored by hand before trusting them. Putting grades and supporting excerpts on the Salesforce record keeps coaching inside tools people already open. Training apps and internal helpers can follow once that core pipeline is stable.