AI agent automating B2B lead qualification from CSV imports to CRM — 1,200 files per month, no human screening.
The agent scores construction-project CSVs against configurable rules and pushes qualified leads into HubSpot. Days of manual screening now take a few hours.
- Days→HrsScreening cycle compressed from days to hours
- AutoThree-outcome decision pipeline, fully autonomous
At a glance
Information was coming in. Insights were not.
OmniTrade Brasil, part of an international group leading the engineering and construction materials market, works on industrial, logistics, and sports projects. Prospecting relied on an external project database: high volume, not ready for sales.
Analysts hand-screened about 1,200 CSV files a month before entering leads in HubSpot: several days per cycle, no consistent criteria.
Excessive manual load. The funnel needed insight, not volume.
The sales team prospected manually with outdated methods: slow, inconsistent across analysts, and impossible to scale without proportional headcount.
1,200 CSV files processed manually per cycle
Each analyst used their own judgement: inconsistent leads, multi-day cycles, no audit trail of why records reached the CRM or not.
Screening, research, and HubSpot onboarding relied on people
Every lead needed a person before reaching the CRM, so more volume meant proportionally more headcount — a structural bottleneck.
An AI agent that automates the entire screening pipeline.
The agent watches Google Drive, filters and scores each record, and sorts leads into three outcomes on criteria the customer edits without engineers.
Inputs
- CSV files in Drive
- Business criteria
- Public company research
Decision engine
AI lead-scoring agent
Outputs
- HubSpot — Hot leads
- HubSpot — UnderReview tag + email
- Discard — logged with reason
- 01Ingestion
Drive monitoring and exclusion filtering
Each new CSV is processed automatically; negative criteria cut the workload before any scoring runs.
- 02AI scoring
Lead scoring with three-outcome decision
Scored 0–100 on CSV fields plus public company research: ≥75 goes to HubSpot, medium is emailed for review, low is discarded.
- 03CRM enrichment
CRM creation, deduplication, and execution logging
Duplicates are caught before insertion, and each lead lands in HubSpot with its score, key criteria and project notes.
Seven capabilities from CSV ingestion to CRM enrichment with execution logs.
- Drive monitoring
- Configurable exclusion filter
- Lead scoring 0–100
- Three-outcome decision
- Duplicate detection
- Automatic CRM note
- Execution log
Faster qualification. Higher-quality leads. Lower technical dependency.
Automated screening, days to hours
The cycle runs autonomously in a few hours, with no analyst time spent on manual filtering.
Qualified leads at the right time
About 10% of processed leads are classified hot and reach the team ready to act, not queued.
Higher-quality leads across channels
Consistent filtering means less noise in the pipeline: the same channels now yield better-qualified opportunities.
A team focused on selling
Repetitive screening moved to the agent; the team’s time now goes to closing deals.
From the team
The problem wasn't volume — it was structure. Once the business criteria had been clearly defined, the agent implemented them consistently. The result wasn't just automation — it was operational predictability.
Stack
Drive → AI → HubSpot, with customer-controlled scoring. No new tools: only systems the team already used.
Building an AI agent for B2B lead qualification?
A no-obligation intro and assessment session to pin down the problem, what success requires, and the right approach for your organisation.
More case studies
