ChiselIQ turns topic intent into submission-ready content packages — researched, drafted, audited and packaged. You review three times and submit once. Never hand a bot the keys to publish.
Built on the stack you already trust
One slash command drives the pipeline; a dashboard gives you a visual cockpit over the whole operation. Everything it produces is an artifact you own.
Finds candidate guest-post and citation sites, then grades each on a 10-dimension CITE rubric. Anything below C is refused for drafting — no exceptions.
Whichever way a topic enters, everything downstream runs the same governed pipeline.
Every claim traces to a source. Unsourceable facts are marked [CITATION NEEDED] — never fabricated.
An automated quality audit grades every draft before it reaches you.
Assembles meta, schema, bio, NAP & outreach email — then tracks which prompts cite the client across AI search.
Research, drafting, auditing and packaging are automated. Every gate is yours.
manual · target · keyword · geo
from topic + policy
H2/H3 structure
GATE 1CORE-EEAT + GEO pass
GATE 2meta · schema · NAP
GATE 3on the target site
HUMANmark-submitted
Every piece of content passes three explicit approvals. ChiselIQ will not run the next stage until you approve the current one — and every approval is written to the tracker.
ChiselIQ tracks which prompts cite the client across model surfaces, aggregates citation share against rivals, and surfaces the prompt gaps to close. No autonomous submits. No account creation. No CAPTCHA solving.
Agencies and operators run ChiselIQ differently — one client or a hundred. We build a plan around your client count, seats and support needs.
Runs on your Claude Code subscription, so there's no metered per-token API bill on core generation. We scope the rest to your workflow.
The pipeline and the dashboard copilot run through the claude CLI on your Claude Code subscription, so there's no metered per-token API bill on core generation — it's covered by the subscription you already have. You can bring your own API keys for optional data connectors (keyword/SERP tools, GSC, GA4).
No — that's the point. ChiselIQ prepares submission-ready packages; a human submits. The only exception is autonomous publishing to the client's own WordPress, and only with explicit confirmation.
Three human approvals — Outline, Draft, Package — are required to move a piece of content forward. They're not skippable; they're the safety mechanism that makes automated content ops viable on real clients.
Partially. The NL copilot can point at an OpenAI-compatible endpoint such as Ollama. The content pipeline is calibrated to Claude's long-form writing and instruction-following, so we recommend a hybrid — keep Claude for content generation, since that's what clients pay for.
Yes. Every brief, outline and draft is an editable artifact you own — edit it directly or use the dashboard's inline editor. Nothing is locked away in an opaque database.
Scans query prompts across model surfaces and record which domains get cited; citation counts aggregate cited URLs by domain. The dashboard shows the last scan's snapshot, not a live per-pageload figure — re-run a scan to refresh. Treat it as a sampled signal, not a monitor.
Yes. Each client is an isolated workspace and data never mixes. Switch between them with a single command; every client keeps its own policies, tracker and artifacts.
Get AI leverage on research and drafting — and never risk a rogue submission on a real client site.