CRM Platform

A CRM platform spanning sales pipeline management, customer records, service work orders, field dispatch, marketing, and an AI sales assistant, designed so service and sales share one customer record instead of living in separate systems.

Businesses that both sell and service customers (installations, work orders, field visits) typically run a CRM for sales and a completely separate system for dispatch and service. Customer context gets lost at the handoff, and sales reps have no visibility into open service issues that affect a renewal conversation.

• Unify sales pipeline and service/dispatch operations around one customer record
• Give dispatch and field teams a workflow built for their operations, not a sales tool repurposed for service
• Automate routine sales tasks (follow-ups, lead qualification) with an embedded AI assistant
• Support marketing campaigns that connect directly to pipeline data instead of a separate tool

A shared customer and account model underpins sales pipeline, marketing, work orders, and dispatch, so any team sees the same history. An AI sales assistant is embedded directly in the pipeline workflow to draft follow-ups and flag at-risk deals, and dispatch operates as its own scheduling and routing workflow against the same account data.

• Modeled customers/accounts as a shared entity referenced by sales, service, and marketing rather than duplicated per module
• Built dispatch/work-order scheduling as an independent workflow engine so field operations aren't bottlenecked by sales-pipeline logic
• Placed the AI sales assistant as an overlay on pipeline data with clearly scoped write actions (drafts, suggestions) rather than autonomous actions on customer records

• Reconciling sales-cycle data models (deals, stages) with service-cycle data models (work orders, dispatch) without forcing one to distort the other
• Building dispatch scheduling that accounts for real-world constraints (technician location, skill, availability) without over-engineering a generic solver
• Keeping the AI assistant's suggestions useful without letting it take actions a rep didn't explicitly approve

• Kept the AI sales assistant strictly suggestion/draft-based at this stage rather than autonomous, based on early user feedback that reps wanted control over customer-facing communication
• Built marketing campaign data as a consumer of the shared pipeline model instead of a bolt-on integration

• Designed the shared customer/account data model across sales, service, and marketing
• Built the dispatch and work-order scheduling workflow
• Directed integration of the AI sales assistant into the pipeline UI

• Gave sales reps visibility into open service issues before renewal conversations for the first time
• Reduced dispatch scheduling time through a purpose-built workflow instead of a repurposed sales tool
• AI-drafted follow-ups reduced time spent on routine pipeline admin

• Sharing one customer record across sales and service required real data-modeling discipline up front, but eliminated an entire category of handoff bugs later
• Users trusted an AI assistant far more once its actions were limited to drafts they approved, rather than autonomous sends