Skip to main content
Talent Intelligence Platform · v3

From 17 generic applications to 5 winning offers

Multi-source candidate consolidation · ANZ market scan · Strategy Mode 3 thinking modes · 17-JD batch apply with Kiwi cover letters · scale-to-N onboarding. Production-ready ADLC plugin · BSL-1.1 license.

⭐ North-Star Metric
% of candidate sessions producing a 5-section consulting-grade Career Strategy Report in <30 min
Target: ≥85% by 2026-Q3 · ≥95% by 2026-Q4 · ≥99.5% by 2027-Q1

Five capabilities. One pipeline.

Every capability has a measurable output, a target metric, and a verifiable acceptance test.

🧬
RQ1
Multi-Source Candidate Consolidation
One profile from 8 sources in <2 min
CV (.docx/.pdf/.md) + LinkedIn PDF + GitHub API + portfolio sites + community signals → canonical YAML with per-field provenance. Eliminates 17× re-parsing across batch applications.
8 sources → 1 SSOT
📊
RQ2
ANZ Market Scan (Top-30)
P25/P50/P75 bands, sector heatmap, 30 company briefs
Brave Search MCP + WebFetch native produce vetted Top-30 NZ/AU enterprise+startup target list with P25/P50/P75 NZD bands per sector. Every band cites source OR marks confidence: low — zero fabricated numbers.
$30-60k NZD negotiation lift
🎯
RQ3
Strategy Mode (3 Modes × Top-3 Roles)
Consulting-grade 6-section DOCX in opus quality
Career Advisor (Strategy Mode, opus override) produces 🧠 Strategic + 🔍 Critical + 🌐 Systems thinking per Top-3 Role with AS-IS→TO-BE LinkedIn + resume rewrites. Cites resume.md evidence per claim.
≥600 words per role · cite-rate ≥95%
🚀
RQ4
17-JD Batch Apply (Top-5 + Breadth-12)
Tailored ATS resumes + Kiwi cover letters in <8 hrs
bin/rank-jds scores 17 SEEK URLs by ATS match × WSJF. Top-5 (≥80%) get full tailoring (resume + 4-paragraph Kiwi cover letter). Breadth-12 get 200-word fallback letters. ATS ≥85 gate per variant.
3-5× conversion lift vs untailored
🌐
RQ5
Scale-to-N Onboarding (5-Q AskUserQuestion)
Zero-touch candidate intake at any volume
Tier-1 capped onboarding: Salary · Mode · Visa · Sector · Size. Emits valid candidate YAML in <90s. Same pipeline reusable for senior IC, early-career, multi-candidate batch manifests.
<90s per candidate intake

How the pipeline runs

Three flows. All composable from the existing talent plugin · zero CrewAI/SerperDev dependency.

HITL: /talent:profile --candidate thanh-nguyen
                         ↓
  commands/profile.md (50 LOC NEW)
                         ↓ delegates
  @research-consolidator (sonnet, 80 LOC NEW)
    ├─ resume-analyzer × 3 (CV.docx, CV.pdf, CV.md)        → resume.json
    ├─ WebFetch Linkedin-Profile.pdf                       → linkedin.json
    ├─ GitHub MCP users/nnthanh101 + topRepos              → github.json
    ├─ WebFetch × 5 demos                                  → demos.json
    │   (cloudops, devops, adlc, analytics, FB group)
    └─ Brave Search MCP "Thanh Nguyen Auckland"            → public.json
                         ↓
  bin/profile-merge (140 LOC NEW)
    ├─ Precedence: resume > LinkedIn > GitHub > portfolio > demos
    ├─ Dedup: Jaro-Winkler ≥ 0.85 (companies / certs / skills)
    ├─ Conflict: surface BLOCKED: per field (no silent picks)
    └─ Provenance: per-field source citation
                         ↓
  skills/multi-source-merge (90 LOC NEW) — jsonschema validate
                         ↓
  tmp/candidates/thanh-nguyen-v2.yaml  (canonical SSOT)

Five input metrics feed the North-Star

Each metric has a measurement method, a Q3 2026 target, and a Q4 2026 target.

Time-to-Report
2026-Q3 target
≤25 min P50
2026-Q4 target
≤18 min P50
Match-Score Precision
2026-Q3 target
≥80%
2026-Q4 target
≥90%
AS-IS / TO-BE Specificity
2026-Q3 target
≥90%
2026-Q4 target
≥99%
Cite-Rate
2026-Q3 target
≥95%
2026-Q4 target
≥99.5%
Candidate CSAT
2026-Q3 target
≥7.5
2026-Q4 target
≥8.5

What four buyers ask. What the page answers.

CTO · CFO · CSO · HR Director. Each gets a 30-second answer with a code-path proof point.

⚙️
CTO
Asks: “Does this reduce hiring engineering-hours per req?”
Yes — 9 hr → 1.5 hr per req (5.5×). Batch-tailor parallelizes resume+cover+ATS across N JDs.
↳ bin/batch-tailor + bin/budget-guard ($1 cap/run)
💰
CFO
Asks: “What's the unit economics? Vendor lock-in?”
$7 LLM/candidate full RQ1-RQ4 run. Zero SaaS subscription. Brave MCP free 2k/mo. SerperDev rejected — no Google Search API spend.
↳ Cost guard (haiku/sonnet/opus split) · MCP profile inventory
🎯
CSO
Asks: “Defensibility vs LinkedIn Coach / BCG advisory?”
3-axis wedge: 5W1H per role (BCG depth) + Kiwi locale (LinkedIn neutral) + ANZ compliance (APRA CPS 234, NZ Privacy Act IPP12). No single incumbent has all 3.
↳ agents/career-advisor.md Strategy Mode · skills/nz-tone-conventions
👥
HR Director
Asks: “Workday/Taleo/Greenhouse compatibility? Compliance?”
JSON Resume v1.0 export ingestible by all 5 ATS. APRA CPS 234 evidence pack via bin/apra-evidence-pack. FCRA disclosures via bin/fcra-disclosure-generator.
↳ 20 bin/ tools · plugin.json userConfig

Quantified ROI

Six metrics. Before pipeline · After pipeline · Lift. Source citations in the blog post.

Metric
Before
After
Lift
Hiring engineering-hours / req
9.0 hr
1.5 hr
5.5×
ATS pass-through (Workday/Greenhouse)
15-25%
60-75%
3.0×
Recruiter screen-to-call rate
8-12%
35-45%
3.5×
Final offer conversion / apply
1.5-3%
8-12%
4.0×
Negotiation leverage (NZD delta)
Baseline
+$30-60k
+15%
LLM spend / candidate full run
n/a
~$7
Quantified

Run the pipeline on your own candidate

Five commands. Real outputs in tmp/applications/. No CrewAI. No SerperDev. No subscription. Cost-bounded by bin/budget-guard.

Read the C-suite blog →View source on GitHub
ADLC Framework v3.7.4 · talent plugin v1.0.0 · adlc.oceansoft.io · BSL-1.1