Skip to content

AI / Automation

AI Sales Automation

Intelligent Lead Qualification & Pipeline Engine

An automated qualification and outreach engine that scores inbound leads, enriches them and routes each to the right owner within minutes.

  • Python
  • OpenAI
  • n8n
  • PostgreSQL
  • GoHighLevel
Abstract diagram of a lead scoring and routing pipeline with branching decision paths

Overview

Project Overview

A revenue team was drowning in inbound volume of wildly uneven quality. We built an engine that ingests every lead from every source, enriches it against public and internal data, scores it against a rubric agreed with the sales leadership, and either routes it to an owner with a briefing or places it into a nurture sequence. Reps stopped triaging and started selling.

Services provided

  • AI Automation
  • Workflow Automation
  • CRM & API Integrations
Client
Confidential — B2B software vendor
Industry
SaaS & Technology
Timeline
8 weeks
Year
2025
Platforms
Automation platform, CRM
Team
2 engineers, 1 delivery lead

Challenge

The Challenge

Lead response time averaged over a day because every enquiry was triaged by hand, and no one could tell which acquisition channels were actually producing revenue.

  • Leads arrived through six disconnected channels with no shared format or deduplication.

  • Manual triage meant the average first response took more than 24 hours.

  • Scoring was intuition-based and varied by whoever happened to pick up the record.

  • Channel spend decisions were made without reliable attribution.

Solution

What We Built

We separated the deterministic work from the judgement work: rules handle routing and thresholds, models handle interpretation of unstructured text.

  • Built a single ingestion layer normalising every source into one validated schema.

  • Used models only for the genuinely fuzzy parts — reading free-text enquiries and summarising company context.

  • Encoded the scoring rubric as versioned configuration so sales leadership can change weightings without a deploy.

  • Added a review queue for borderline scores rather than forcing the system to guess.

  • Instrumented the full funnel so channel performance is measured against closed revenue, not lead count.

Capabilities

Key Features

What AI Sales Automation does in day-to-day use.

  • Unified lead ingestion

    Forms, ads, chat, email and partner feeds normalised into a single schema with deduplication on arrival.

  • Automatic enrichment

    Company size, sector and technology signals appended before a human ever opens the record.

  • Rubric-based scoring

    Leads scored against explicit, versioned criteria so the reasoning behind a score is always inspectable.

  • Ownership routing

    Territory, capacity and specialism rules assign each lead to the right rep with a written briefing.

  • Adaptive nurture

    Below-threshold leads enter a sequence that re-scores them as new signals arrive.

  • Pipeline reporting

    Source quality, conversion by score band and response-time tracking in one view.

Stack

Technology Stack

The tools this system runs on, grouped by the role they play.

  • Automation

    • n8n
    • Make

  • Backend

    • Python
    • FastAPI
    • Celery

  • AI

    • OpenAI
    • Claude
    • Text embeddings

  • CRM

    • GoHighLevel
    • REST integrations

  • Data

    • PostgreSQL
    • Redis

  • Infrastructure

    • Docker
    • GitHub Actions

Architecture

How It Works

The path a single request takes through the system, end to end.

    INGEST01

    Capture & normalise

    Every source lands in one validated schema, deduplicated on arrival.

    ENRICH02

    Append context

    Firmographic and behavioural signals attached automatically.

    SCORE03

    Evaluate

    Rubric scoring with model-read free text and an inspectable rationale.

    ROUTE04

    Assign or nurture

    Above threshold goes to an owner with a briefing; below enters nurture.

    MEASURE05

    Close the loop

    Outcomes feed back into channel attribution and rubric tuning.

Outcome

Results

What changed for the business after launch.

  • < 5 min

    Median first response

    Down from more than a day under manual triage.

  • 3.4×

    More qualified conversations

    Per rep per week, with no increase in headcount.

  • 12 hrs

    Manual triage removed weekly

    Time returned to the sales team across the pipeline.

  • 6 → 1

    Lead sources unified

    One schema, one queue, one source of truth.

Interface

A Closer Look

Screens from the delivered system.

  • Lead scoring rubric configuration screen
    Versioned scoring rubric — editable without a deployment.
  • Routing rules and territory assignment view
    Routing rules combining territory, capacity and specialism.
  • Funnel attribution dashboard
    Channel attribution measured against closed revenue.

More work

Related Projects

Other systems built on similar foundations.

View all projects
  • Abstract diagram of a multi-agent voice routing system with connected conversation nodes

    AI / Voice / Multi-Agent

    VeriVoice

    An intelligent multi-agent voice platform designed to automate business conversations and workflows.

    • React
    • Next.js
    • Node.js
    • AI
    • +1

    View Case Study

  • Abstract scheduling diagram showing appointment slots and voice interaction paths

    AI / Voice / Automation

    MediSync Scheduling Assistant

    A voice and messaging assistant that handles appointment booking, rescheduling and reminders for a multi-site clinic group.

    • Twilio
    • ElevenLabs
    • OpenAI
    • Node.js
    • +1

    View Case Study

Next step

Want a System Like This?

Tell us what your business needs to automate and we'll map out a practical build.

Book a Free ConsultationContact Us

hello@devrox.comWe reply to every enquiry within one business day.