Hashmeta AI-SEO — S$350,000 incremental revenue from one multi-agent system.
How we built a multi-agent SEO system that ranked #1 on 12+ competitive Singapore terms in 45 days, landed the first paid client in 60, and generated S$350,000 in incremental revenue with no ad spend.
| Client | Hashmeta (internal) |
|---|---|
| Role | Partner & CTO; system architect |
| Dates | 2024–2025 (live and compounding) |
| Outcome | S$350k incremental revenue · 5M organic clicks/year · 12+ #1 rankings in 45 days · first paid client in 60 days |
Hashmeta's SEO leads had stagnated. The agency was ranking on a handful of branded terms but losing on the head terms that mattered for new business — “AI Agency Singapore,” “AI SEO Agency Singapore,” “AI Marketing Agency Singapore.” The content team couldn't sustain agency-scale publishing on competitive terms with editorial standards intact, and ad spend was already flat.
The brief was simple, the constraint wasn't:
Publish at agency-scale on competitive head terms — without dropping editorial standards, and without hiring five more writers.
A multi-agent SEO system architected around the Expert in the Loop model:
- Strategist agent — keyword research, search-intent classification, content-brief generation
- Writer agents — long-form drafts in the Hashmeta voice, with citation discipline
- Editor agent — fact-checking, internal linking, schema markup
- Publisher agent — final formatting, image selection, publish scheduling
- Reviewer in the loop — me, with hard stop-points at brief approval and final-draft approval
The system was never autonomous. It was leveraged. One senior pass per article, ten articles per week. That's the throughput multiplier.
- Models: Claude (orchestration + writing)
- Skills: OpenClaw skill catalogue — keyword research, brief generation, fact-check, internal-linking, schema
- Orchestration: custom Python harness chaining skills with stop-points
- Publish: WordPress with custom schema deployment
- Metrics loop: Google Search Console + Ahrefs feeding back into the strategist agent
- The editor agent was the bottleneck, not the writers. We over-invested in writer-agent quality early; the leverage was in fact-check + linking discipline downstream.
- Brief quality dictated everything. When briefs drifted, articles drifted; when briefs were tight, the senior pass took 5 minutes per article instead of 25.
- Ranking velocity surprised even us. 45 days to #1 on terms we'd been losing for two years was not the timeline we'd promised the partners.
- Earlier rigor on the editor agent — we let drafts drift in week 2.
- Brief library before brief automation — curate 50 briefs by hand before automating brief generation.
- Day-one metrics loop — instrument the rankings → content feedback loop on day one, not week three.
Pillar: AI Marketing Frameworks & SEO/GEO
Spoke: AI SEO Blueprint: how to drive 5M organic clicks/year — the public version of the playbook behind this case study.
Want a system like this for your business?