Newstools for publishers
Newsroom assistants
Discover, research, draft, and verify in one workflow, with the newsroom in control.
Let's talk about your project
Proposals, not decisions
Every headline, tag, or metadata field arrives with context and confidence; the editor accepts, corrects, or rejects.
Inside your CMS
Connectors for WordPress, Strapi, and Airtable, plus webhooks and n8n, without switching tools.
Tags with provenance
Enrichment from Wikidata, Knowledge Graph, and IAB taxonomies, with provenance and review status recorded.
One story, several assistants, one editor
A person decides at this step
- Story in the CMSA new or archived article comes in through a connector or webhook with its text and current tags.
- AnalysisAgents extract entities and claims and cross-reference signals from Wikidata, Knowledge Graph, and IAB taxonomies.
- ProposalsHeadline, summary, metadata, citations, translated version, and tags, each with confidence and provenance.
- Editor's decisionAccepts, corrects, or rejects each proposal from their own tool; nothing is applied without that decision.
- AppliedWhat was accepted goes back to the CMS in its fields and taxonomies; who decided what is recorded.
- TracesHistory, prompts, models, and costs per story, in the dashboard organized by teams.
Nothing is applied to the CMS without the editor's decision.
What it solves
Every story drags along tasks that are not writing: an alternative headline, a summary, metadata, citations, a version in another language, the CMS tags. With catalogs of thousands of tags, classification becomes inconsistent and the archive loses value for search, navigation, and analysis.
We put AI assistants at the point where the newsroom needs them, inside its workflow. Every proposal arrives with context and confidence, and an editor accepts, corrects, or rejects it.
How we work
We start from a workspace that brings together discovery, research, drafting, verification, and publication preparation. Its modules can operate separately or in a connected workflow that retains sources, results, and editorial review; deployment is adapted to the newsroom, its AI models, and its systems.
For editing and adaptation, the assistants prepare headlines, summaries, SEO metadata, and citations; translate and adapt content using style guides; and review claims and references with evidence. Analysis of AI-generated text signals is indicative only.
For taxonomy, we can cross-reference signals from Wikidata, Knowledge Graph, IAB taxonomies, and content analysis, and we record provenance, confidence, and review status for every tag proposal. The resulting semantic base supports search, navigation, and content analysis.
What we deliver
- Assistants integrated with your CMS through connectors for WordPress, Strapi, and Airtable, plus webhooks and n8n, with fields, taxonomies, media, and delivery rules configured for each destination.
- A reviewable tag enrichment workflow, with entities, categories, and semantic relationships, and a record of who accepted, corrected, or rejected each proposal.
- Traceability of history, workflow artifacts, prompts, models, token consumption, and costs, and a dashboard that organizes work by teams and organizations.
- Publication conditioned, if you want it, on quality checks and editorial approval.
Commercial segmentation and propensity models require their own projects and validation: they are not automatic results of enrichment.
Who it is for
Newsrooms that want AI support on specific tasks without changing their CMS or the way they work, and publishing groups with large tag catalogs that need to organize them with judgment and keep control over every change. Specific capabilities and permissions are agreed during deployment.
What we have already built
Newstools Suite is the workspace these assistants come from: discovery and planning, multi-source research and generation, editing and adaptation tools, connected publishing, and traceability. It organizes planning, drafting, refinement, verification, evaluation, and review before export, and it is deployed module by module according to what your newsroom needs.