AI Sales Automation CRM

A lead-management system designed to automate prospect capture, qualification, and follow-up, with pipeline visibility for the whole sales process. The project is in development, starting from the lead-source and qualification model.

Growing businesses often collect leads from several channels, ads, forms, referrals, without a shared system to qualify and follow up on them consistently.

Leads arrive from different sources with no consistent qualification step, follow-up depends on individual memory, and there is no single view of pipeline value or conversion activity.

A CRM layer that ingests leads from defined sources, applies qualification rules automatically, and triggers scored, staged follow-up sequences. AI-assisted drafting helps prepare follow-up emails, while a dashboard shows pipeline value, conversion rate, and response activity.

• Small sales teams managing leads from multiple channels
• Founders handling sales personally without dedicated CRM support
• Agencies that need a shared pipeline view across client accounts

1. 1

   A lead enters the system from a form, ad platform, or manual entry

2. 2

   Qualification rules score the lead and assign a pipeline stage

3. 3

   A follow-up sequence is triggered with AI-assisted email drafts queued for review

4. 4

   Reminders surface leads that need the next action

5. 5

   Pipeline dashboards update to reflect stage, value, and response activity

• Lead source mapping and qualification rules
• AI-assisted email drafting and lead scoring
• Reminders and next-action recommendations
• Pipeline dashboard views

A Next.js application provides the pipeline dashboard and lead views, backed by a Python service handling scoring logic and OpenAI-assisted drafting. PostgreSQL stores lead, stage, and activity records, with n8n handling scheduled follow-up triggers and integrations with lead sources.

• New leads are scored against defined qualification criteria on intake
• Stage changes trigger the corresponding follow-up sequence and reminder schedule
• Drafted follow-up emails are queued for human review before sending

• Lead and contact data access is scoped to authenticated users
• AI drafting calls only send the minimum context needed to produce a useful draft
• Credentials for connected lead sources are stored as environment variables

• Designing qualification rules general enough to apply across different lead sources
• Balancing automated follow-up cadence against the risk of over-contacting a prospect
• Keeping AI-drafted messages accurate to the actual offer and pricing

• Gives every lead a consistent qualification and follow-up path
• Reduces reliance on memory for who to contact next
• Provides a shared pipeline view for reporting and prioritization

• The system is in development and has not processed live production leads
• Scoring rules are based on a general model and would need tuning per business
• Dashboard views are still being defined

• Add configurable scoring weights per business type
• Introduce two-way sync with common ad-platform lead forms
• Add a forecasting view based on pipeline stage history