How Tezeract Built COAS, a Candidate Outreach Automation System That Gives Recruiters Full Control Over AI-Driven Outreach

Impact

Project Overview

Recruitment agencies operating across European markets face a problem that compounds at scale: too many CVs in too many languages, too many live job listings to manually track, and outreach that cannot grow without losing the recruiter’s judgment at the point of action.

Ben Brabers, founder of Brabers, a UK-based recruitment firm operating across English, Dutch, and German-speaking markets, needed more than a tool to speed things up. He needed a platform that could parse multilingual CVs, match candidates to live job openings with transparent, auditable reasoning, and send recruiter-approved outreach without ever removing human control from the loop.

Tezeract designed and built COAS from scratch: a fully custom candidate outreach automation system with AI CV parsing, an eight-step explainable matching engine, multilingual outreach drafting, automated follow-up sequences, reply intent classification, and a GDPR/CCPA-compliant campaign dashboard. Delivered across four milestones in approximately eleven weeks.

What Changed

Recruiters at Brabers who previously spent hours formatting CVs, manually drafting outreach in multiple languages, and chasing non-responders with no tracking system can now upload a CV in any supported language, review an AI-ranked match list with step-by-step reasoning, approve outreach in one click, and monitor every campaign from a single dashboard, with full compliance logging throughout.

Customer Profile

Why This Matters for Buyers Like You

If you run a recruitment agency or talent platform that handles cross-border placements, multilingual CV pipelines, or high-volume outreach campaigns, COAS is a direct proof of concept. The challenge Brabers faced is not specific to one firm. Agencies across Europe deal with the same combination of language barriers, opaque ATS matching scores, and outreach that cannot scale without sacrificing recruiter oversight.

What COAS demonstrates is that automation and human control are not opposites. It is possible to build an AI recruitment tool that handles the volume while keeping every outbound decision in the recruiter’s hands. If your team is spending hours on CV formatting, manual translation, or guessing why a candidate didn’t match a role, this case study shows exactly what a different approach looks like.

When Recruitment Runs on Manual Work in a Market That Won't Wait

Primary Problem

The core problem was scale and control at the same time. Brabers needed a platform that could handle multilingual CV intake, match candidates to live job listings daily, and send compliant personalized outreach at volume, but every AI action had to route through the recruiter before it became an outbound action. No black-box scoring. No emails sent without explicit approval.

Building an AI candidate matching platform that is both automated and auditable is a harder engineering problem than building one that simply automates. COAS had to do both.

Secondary Challenges

CVs arrived in English, Dutch, and German with no consistent way to normalize and compare candidates across languages.

Candidate-to-job matching produced no explanation of why a match succeeded or failed at each step.

Outreach emails had to be drafted, translated, and manually sent per campaign, with no structured follow-up system in place.

There was no mechanism to detect when a company had already been contacted, creating duplicate outreach risk and sender reputation damage.

Campaign performance had no centralized view: who replied, who didn’t, and whether intent was positive or negative.

GDPR and CCPA required audit trails that the existing process did not produce.

Still Spending Hours on Manual Candidate Outreach?

If multilingual CV reviews, manual matching, and repetitive outreach are slowing your recruiters down, it’s time to build a system designed around your workflow. See how AI can automate the heavy lifting while keeping your team in full control of every decision.

What Slowed Down Operations and Triggered the Need for Immediate Change

Previous Solutions Tried

Generic ATS platforms that lacked multilingual CV support and returned opaque match scores. Manual email workflows in Gmail and Outlook that could not scale beyond a handful of active campaigns. Off-the-shelf recruitment tools that offered no explainability, no language handling, and no recruiter approval gates between AI suggestion and outbound action.

Business Impact

The manual process created a hard ceiling on how many candidates Brabers could actively work with at any given time. Recruiters spent the bulk of their working hours on formatting, translating, and chasing follow-ups rather than on judgment-level work. Duplicate outreach to the same company damaged sender reputation. Compliance risk was growing alongside outreach volume. The business model could not scale without a structural change to the process.

Urgency Factors

Why Tezeract

Ben Brabers came to the engagement with deep recruitment expertise and one non-negotiable requirement: the system had to be explainable. Every match decision had to be auditable by the recruiter. Off-the-shelf recruitment automation software development tools returned scores. Tezeract returned reasoning.

Why Tezeract Won the Evaluation

Building COAS required a specific combination of capabilities that most development teams cannot deliver together:

AI-powered multilingual CV parsing: Extracting structured fields from PDFs and DOCXs in English, Dutch, and German, normalized to English for consistent downstream processing.
Explainable multi-step matching: Eight sequential filters with written pass/fail reasoning at every step, not a single opaque score.
Recruiter-in-the-loop architecture: Every AI output, parsed CV, match suggestion, drafted email, follow-up, routes through explicit recruiter approval before any outbound action is taken.
Full-stack outreach pipeline: CV parsing, matching, email drafting, translation, follow-up scheduling, reply classification, and campaign analytics, all in one platform.
Compliance-grade logging: GDPR and CCPA audit trails built into the platform, not added as an afterthought.
Milestone-based delivery: Four sequenced phases, each closing with a deployed working demo and a structured feedback round.

Tezeract’s track record in AI automation services for B2B platforms, combined with a delivery model that locked architecture and API contracts before feature development began, gave Ben Brabers confidence that COAS would be delivered to spec and on schedule.

How Tezeract Built a Custom Candidate Outreach Automation System That Keeps the Recruiter in Control at Every Step

How Nava AI Works

A recruiter uploads CVs in any supported language. COAS detects the file type, extracts and normalizes the content to English, and pulls structured fields automatically. Two versions are generated from the same parse: a branded CV and a fully anonymized one. The recruiter reviews, edits if needed, and approves before any candidate record is created.

From there, a daily scheduler scrapes fresh LinkedIn job listings, enriches each company via Apollo, and runs the matching engine across all candidates. Eight sequential filters apply with written reasoning at every step, giving the recruiter a ranked match list that explains exactly why each candidate passed or failed.

Approved matches trigger an AI-drafted outreach email in the recruiter’s chosen language, with the anonymized CV attached, reviewed and approved before anything sends. Replies are classified as positive or negative and everything logs to the dashboard for compliance and performance tracking.

Key Capabilities Built

Multilingual CV Parsing with English Normalization

Accepts PDF, DOC, and DOCX in English, Dutch, and German. Language detection is automatic. All CVs normalize to English so candidates across all three languages are matched against the same job pool on equal terms.

Dual CV Output with Version History

Every parse produces a branded templated CV and an anonymized CV with sensitive identifiers stripped. Recruiter edits to either version sync across both under a single tracked version history. Industry exclusion controls are set per candidate.

Eight-Step Explainable Matching Engine

Filters run sequentially: agency exclusion, work mode preference, time zone, industry, role alignment, language fluency, location, and overall AI score against a configurable threshold. Every filter produces a written pass or fail with a specific logged reason. Rejection reasons feed back into model improvement.

AI-Drafted Multilingual Outreach

Once a match is approved, COAS drafts the outreach email in the recruiter’s chosen language with the anonymized CV attached. The recruiter reviews and approves the initial email and all follow-up drafts before any send.

Automated Follow-Up Sequences with Override Controls

Follow-ups go only to genuine non-responders on the configured schedule. Recruiters can stop any follow-up manually at any point. Duplicate-outreach warnings fire if the same company is targeted in a new campaign.

Reply Intent Classification

Incoming replies are automatically classified as positive or negative. Positive replies surface immediately in the recruiter’s view. No reply goes untracked.

Campaign Dashboard with GDPR/CCPA Audit Trails

Every action is logged with timestamps, reasons, and status changes. The dashboard provides Kanban views, per-campaign analytics, pipeline metrics, and a full compliance audit trail.

Need These Features Built for Your Recruitment Business?

From AI CV parsing and explainable matching to multilingual outreach and compliance tracking, we can build a recruitment platform tailored to your exact process and market requirements.

Phases wise Deployment

Tezeract delivered COAS across four sequenced milestones, each closing with a deployed working demo and a structured client feedback round. Architecture and API contracts were locked in Milestone 1 before any feature development began, a deliberate decision that prevented rework in every phase that followed.

Phase 1: Foundation, CV Management and Parsing

Architecture, ERD, and API contracts locked before feature development began. UI/UX design with live client feedback cycles completed. Apify LinkedIn API tested. Database schema, authentication, onboarding flows, and candidate management module built. Full CV extraction API deployed: file-type detection, language detection, English normalization, and structured field extraction. AWS deployment infrastructure established.

Key Milestone: End-to-end CV ingestion working from upload to structured candidate record. Architecture locked and validated with client.

Phase 2: Matching Engine and Campaign System

Team member management and invitation flows completed. LinkedIn job fetcher with Apollo enrichment deployed and running on daily schedule. Multi-step explainable matching algorithm built with rejection-reason logging active from day one. Campaign creation, email translation, email styling APIs, and daily matching scheduler completed.

Key Milestone: Matching engine live with per-step written reasoning across all eight filters. Campaign creation and multilingual email drafting functional.

Phase 3: Campaign Tracking, Intent Classification and Dashboard

Campaign tracking workflows built: active, stopped/paused, historical, reply, Kanban, and analytics views. Main admin dashboard completed. CV template settings deployed. Reply intent classification API live, enabling positive/negative categorization of all incoming responses.

Key Milestone: Full campaign lifecycle trackable from a single dashboard. Intent classification running on all incoming replies. GDPR/CCPA audit trail logging active.

Phase 4: Client Feedback, Testing and Final Delivery

Client feedback from Phase 3 demo implemented. UX refinements applied across the interface. Regression testing, performance testing, and acceptable load testing completed. Final demo delivered to Brabers. Production deployment signed off and live.

Key Milestone: Full platform live in production. All four milestones delivered on schedule across approximately eleven weeks.

Obstacles Countered and Resolved

Explore the key challenges faced during the project and how our team addressed them with practical solutions.

Obtacles

Resolution

The Results

COAS delivered a complete, recruiter-controlled outreach loop across multilingual candidate pools and live LinkedIn job listings within eleven weeks.

Every step that previously required manual input, CV formatting, language normalization, match evaluation, email drafting, follow-up tracking, is now handled by the platform, with recruiter approval gates preserving full human oversight.

What these results mean for different stakeholders:

For Recruitment Agency Founders

Replace manual CV formatting, match evaluation, and email drafting with a fully automated pipeline that still requires your approval at every outbound step.

Get written match reasoning at every filter so you can defend every placement decision to clients and candidates.

GDPR and CCPA audit trails reduce compliance exposure as outreach volume grows.

For Senior Recruiters

Work through a daily AI-ranked match list instead of manually comparing candidates to job listings one by one.

Draft, translate, and approve outreach in three languages from a single interface without switching tools.

Positive replies surface immediately; no warm response gets buried in a general inbox.

For Recruitment Operations

Every action taken by the system is logged with timestamps, reasons, and status.

Anonymized CVs are generated automatically, reducing candidate PII exposure in outreach.

Duplicate-outreach controls and a full audit trail keep campaigns clean and defensible.

Looking for Explainable AI Instead of Black-Box Recruitment Tools?

We help recruitment firms build AI systems that provide transparency, auditability, and full human oversight. No hidden scoring models. No unexplained decisions.

Building COAS With a Purpose-Built AI Recruitment Technology Stack

GPT4

Apify

Pilotter

Apollo

Postmark

MongoDB

AWS

Layer

Tools

Role

AI and LLM

GPT-4o

CV parsing, field extraction, anonymization, outreach email drafting

Language Processing

Custom pipeline

Language detection (EN/NL/DE), English normalization for consistent downstream matching

Matching Engine

Custom rule-based + AI-scored

Eight-step sequential filter pipeline with configurable thresholds and rejection-reason logging

Intent Classification

Custom model

Reply classification: positive vs. negative intent

Job Scraping

Apify, Pilotter

Daily LinkedIn job and company data scraping

Contact Enrichment

Apollo

Decision-maker contact and company enrichment

Email Delivery and Tracking

Postmark

Outreach send, reply tracking, bounce monitoring

Database

MongoDB

Candidate records, job data, campaign data, audit logs

Cloud Infrastructure

AWS

Containerized, modular deployment with independent service scaling

Compliance

Built-in audit trail

GDPR/CCPA logging across all platform actions

COAS’s full-stack architecture was built using our business process automation services, combining LLM-powered parsing, custom matching logic, and multilingual NLP into a single production-grade platform.

Key Features

Multilingual CV Parsing with Dual Output Generation

COAS accepts CVs in English, Dutch, and German, detects the language automatically, and normalizes all content to English for consistent downstream matching. From a single parse, the platform generates two CV versions: a branded templated version with the agency’s header and footer, and a fully anonymized version with contact details and sensitive identifiers stripped.

Eight-Step Explainable Matching Engine

Unlike tools that return a single opaque score, COAS runs each candidate through eight sequential filters: agency exclusion, work mode preference, time zone, industry, role alignment, language fluency, location, and an overall AI score against a configurable threshold. Every filter produces a written pass or fail with the specific reason logged. Recruiters see exactly why a candidate matched or didn’t match a role.

AI-Drafted Multilingual Outreach with Anonymized CV Attachment

Once a recruiter approves a candidate-job match, COAS drafts a personalized outreach email in the recruiter’s chosen language: English, Dutch, or German. The anonymized CV is attached automatically. The recruiter reviews and approves both the initial email and every scheduled follow-up draft before anything is sent.

Automated Follow-Up Sequences with Override Controls

COAS sends follow-up emails on a configured schedule to contacts that haven’t responded. The system tracks response status per contact and only follows up genuine non-responders. Recruiters can stop any follow-up sequence manually at any point. If the same company appears in a new campaign that has already received outreach, a duplicate-outreach warning fires before the recruiter proceeds, protecting sender reputation across all active campaigns.

Campaign Dashboard with GDPR/CCPA Compliance Tracking

Every action in the system is logged: match decisions and rejection reasons, outreach sent, follow-ups triggered, replies received, and status changes. The admin dashboard provides Kanban views, per-campaign analytics, pipeline metrics, and a compliance-grade audit trail that satisfies GDPR and CCPA requirements. Positive replies surface immediately in the recruiter’s view so no warm response gets missed.

How Candidate outreach automation system Helps

Cross-Border European Recruitment Agencies

Agencies placing candidates across UK, Netherlands, and Germany can use COAS to handle the full multilingual CV pipeline without a separate tool for each market. Language normalization ensures Dutch and German candidates are matched against the same job pool on equal terms, and outreach goes out in the recipient’s preferred language automatically.

High-Volume Outreach Campaign Management

Firms running simultaneous outreach campaigns to multiple companies can use COAS to manage follow-up sequences, track response rates, and classify reply intent across all active campaigns from a single dashboard. The duplicate-outreach warning system prevents sending to the same company twice, keeping sender reputation intact at scale.

Compliance-Sensitive Recruitment Operations

Any agency operating under GDPR or CCPA requirements can use COAS to ensure anonymized CVs go out in all outreach, all candidate data handling is logged with timestamps and reasons, and the full audit trail is accessible at any point. Compliance is built into the platform and runs automatically alongside every campaign action.

Recruitment Platforms Requiring Explainable AI Decisions

Any B2B recruitment SaaS that needs to show clients or regulators why a candidate was or wasn’t matched to a role can replicate the COAS matching architecture. The sequential filter model with logged rejection reasons maps directly to client reporting, reducing the time spent justifying placement decisions to zero.

B2B Outreach Platforms With Human Approval Gates

The COAS architecture is not limited to recruitment. Any B2B outreach platform that requires a human approval step between AI-generated suggestions and live outbound action can use the same recruiter-in-the-loop model. AI drafting, human approval, automated follow-up, and reply classification apply cleanly to sales development, partnership outreach, and account-based marketing programs.

Ready to Build Your Own Candidate Outreach Automation System?

COAS took Brabers from a manual, language-constrained outreach process to a fully automated, compliant, multilingual recruitment platform in eleven weeks. If your agency or platform is dealing with the same bottlenecks, our team has already solved the hard problems: multilingual parsing, explainable matching, recruiter approval gates, and compliance logging that scales with your campaigns.

As a full-service AI development company, Tezeract has delivered AI recruitment automation systems, custom NLP pipelines, and complex B2B SaaS platforms for clients across Europe, the US, and the Middle East. Whether you need a candidate outreach automation system built from scratch or want to add explainable matching and automated follow-up to an existing platform, we’ll map out the architecture on your first call.

Want to understand how we structure delivery before you commit? Explore our AI development process to see how we go from architecture to production.

Frequently Asked Questions

What is a candidate outreach automation system?

A candidate outreach automation system is a platform that automates the end-to-end process of matching candidates to job openings, drafting personalized outreach emails, scheduling follow-up sequences, tracking replies, and logging all campaign activity for compliance. COAS handles all of these steps while keeping a recruiter approval gate at every critical action, so the human stays in control of what goes out and when.

How does COAS differ from a standard ATS?

A standard ATS stores candidate records and tracks pipeline stages. COAS actively matches candidates to live job listings daily, drafts outreach in multiple languages, classifies incoming replies, and runs follow-up sequences automatically. It also adds explainability that most ATS platforms don’t offer: recruiters see exactly why a candidate passed or failed each of the eight matching filters, not just a score.

Can COAS handle CVs in languages other than English?

Yes. COAS supports CV uploads in English, Dutch, and German. The system detects the language automatically and normalizes all content to English before matching, so candidates across all three languages are evaluated against the same job pool consistently. Unsupported languages return a clear error so recruiters always know the status of each upload.

What makes the COAS matching engine different from standard AI recruitment software?

Most AI recruitment software returns a single match score with no explanation. COAS runs eight sequential filters and produces a written pass or fail with a specific reason at every step. Recruiters can see exactly where and why a candidate didn’t match a particular role. Those rejection reasons are also logged automatically to improve the model over time, which most tools don’t do.

How does COAS ensure GDPR and CCPA compliance?

Every action in COAS is logged with timestamps and reasons: match decisions, outreach sent, follow-ups triggered, replies received, and status changes. Candidate PII is stripped automatically in the anonymized CV before it is attached to any outreach. Audit trails are always accessible and are built into the platform, not dependent on a third-party integration.

How long does it take to build a system like COAS?

Tezeract delivered COAS in approximately eleven weeks across four milestones. The timeline depends on the features required, the number of integration points, and how customized the matching logic needs to be. Locking architecture and API contracts in the first milestone is what keeps the rest of the build on schedule. You can read more about how we structure delivery on our AI development process page.

What integrations does COAS use?

COAS integrates with LinkedIn via Apify and Pilotter for daily job scraping, Apollo for company and decision-maker contact enrichment, and Postmark for email delivery and reply tracking. AWS provides the containerized cloud infrastructure. MongoDB handles the database layer. The modular architecture means additional integrations such as CRM systems or Slack notifications can be added without rebuilding core components.

Can Tezeract build a similar system for sales outreach or B2B lead generation?

Yes. The core COAS architecture, AI-drafted outreach, human approval gates, automated follow-up sequences, reply classification, and a compliance-grade dashboard, applies directly to sales development and account-based outreach platforms. If you are building in this space, our AI automation services are the right starting point for scoping the build.

How Tezeract Built COAS, a Candidate Outreach Automation System That Gives Recruiters Full Control Over AI-Driven Outreach

Impact

3

8

11 Weeks

Project Overview

What Changed

Customer Profile

Why This Matters for Buyers Like You

When Recruitment Runs on Manual Work in a Market That Won't Wait

Primary Problem

Secondary Challenges

Still Spending Hours on Manual Candidate Outreach?

What Slowed Down Operations and Triggered the Need for Immediate Change

Previous Solutions Tried

Business Impact

Urgency Factors

Why Tezeract

Why Tezeract Won the Evaluation

How Tezeract Built a Custom Candidate Outreach Automation System That Keeps the Recruiter in Control at Every Step

How Nava AI Works

Key Capabilities Built

Multilingual CV Parsing with English Normalization

Dual CV Output with Version History

Eight-Step Explainable Matching Engine

AI-Drafted Multilingual Outreach

Automated Follow-Up Sequences with Override Controls

Reply Intent Classification

Campaign Dashboard with GDPR/CCPA Audit Trails

Need These Features Built for Your Recruitment Business?

Phases wise Deployment

Phase 1: Foundation, CV Management and Parsing

Phase 2: Matching Engine and Campaign System

Phase 3: Campaign Tracking, Intent Classification and Dashboard

Phase 4: Client Feedback, Testing and Final Delivery

Obstacles Countered and Resolved

The Results

3

8

11 Weeks

For Recruitment Agency Founders

For Senior Recruiters

For Recruitment Operations

Looking for Explainable AI Instead of Black-Box Recruitment Tools?

Building COAS With a Purpose-Built AI Recruitment Technology Stack

Layer

Tools

Role

Compliance

Built-in audit trail

Key Features

Multilingual CV Parsing with Dual Output Generation

Eight-Step Explainable Matching Engine

AI-Drafted Multilingual Outreach with Anonymized CV Attachment

Automated Follow-Up Sequences with Override Controls

Campaign Dashboard with GDPR/CCPA Compliance Tracking

How Candidate outreach automation system Helps

Cross-Border European Recruitment Agencies

High-Volume Outreach Campaign Management

Compliance-Sensitive Recruitment Operations

Recruitment Platforms Requiring Explainable AI Decisions

B2B Outreach Platforms With Human Approval Gates

Ready to Build Your Own Candidate Outreach Automation System?

Frequently Asked Questions

Services

Industries

Portfolio

Quick Links