From source to decision, with traceability at every step

The journey, step by step

1. 1

   Define what to monitor

   Create a channel for each monitoring effort and tell it what matters to you.

   • Organizations with administrator, editor and reader roles: anyone can create their account and their organization, and the rest of the team joins by invitation.
   • Channels with rules: languages, countries and keywords to include or exclude.
   • Watchlists by company, person, regulation, technology, place or topic, with synonyms and exclusions.
   • Assisted suggestion of terms and synonyms for each watchlist: you decide which ones to add.
   • Groups to organize channels and people.

2. 2

   Connect your sources

   Assign sources to each channel, one by one or from a package.

   • Eight connectors: RSS and Atom, sitemaps, page changes, email newsletters, RIS, CSV and BibTeX import, document upload, arXiv and OpenAlex.
   • Test before saving: you see the five most recent items the source would bring in.
   • Feed builder: you give an address without RSS and the platform proposes how to follow it.
   • Scheduling by interval or by cron expression, a status dashboard and automatic pausing after repeated failures.

3. 3

   Clean capture, without duplicates

   What comes in is normalized before it reaches your channel.

   • Extraction of the main text and language detection.
   • Exact duplicates are not stored twice, and near-duplicates are grouped under a single item.
   • A copy of the original content for each version, with its address and its capture date.
   • Full-text search with exact phrases, exclusions and language-aware lemmatization.

4. 4

   AI enrichment

   Each item gains context to separate the signal from the noise.

   • Mentioned entities (people, organizations, places and regulations), their link to your watchlists and a view to browse them.
   • Relevance from 0 to 1, with an explanation you can check against the item. Your organization’s ratings fine-tune it within a bounded margin, and the explanation says so.
   • Automatic tags, which you can complete or correct with manual tags.
   • Sentiment toward the entities you follow in a watchlist.
   • Grouping by story and by topic: a news item replicated across ten outlets reads as one story with ten sources, and topics bring together broader, longer-lasting subjects.

5. 5

   Agents that write with citations

   The agents work on your corpus, not on general knowledge.

   • Summary of a piece of content or of a story, and key messages of a channel over a period, with their coverage in plain view: “based on 8 of 23 pieces of content”.
   • Natural-language questions about what has been collected, scoped by channel and by dates, with a history of what was asked.
   • Rating of each item as “interesting” or “not interesting”, and recommendation of content based on those ratings and of sources you could add.
   • Each output references the items it relies on and shows the quoted excerpt; without evidence, the answer is “not enough data”.

6. 6

   Deliverables

   The information reaches the people who decide, in the format they use.

   • Rule-based alerts: words, an entity, a watchlist, a minimum relevance or a spike in publications. Immediate, daily or weekly, in your organization’s time zone.
   • Alerts reach the members of the organization by email and the Telegram chats you register. The same item never arrives twice, there are limits against excessive notices and each person can mute a rule.
   • Newsletters with templates: automatic sections (featured items, stories and key messages with citations) and text from the team. A person reviews and edits them before sending, with preview, test send, scheduling and recipients by group. They reach the members of the organization.
   • Trends dashboard: volume over time, share of voice, entity map and comparison between competitors.
   • Personal and shared collections to save items, and the list of your ratings.

Coming soon

• Reports in PDF, Word and PowerPoint.
• Public API and webhooks.
• Predictive alerts and forecasting: today’s spike alerts compare against the recent average; they do not predict.
• Sending newsletters and alerts to addresses outside the organization.
• Dedicated connectors for regulation, tenders, patents and economic data, public social media, and audio and video.
• Single sign-on and white label.

Frequently asked questions

What does the platform do?

It organizes your team into organizations with permissions and monitors what matters to you with channels and watchlists. It captures what is published through eight connectors, without duplicates and with search, and applies AI enrichment: entities, explained relevance, tags, stories and topics. Its agents write summaries and key messages with citations, answer questions with citations, collect your ratings and make recommendations. Everything reaches the decision makers as rule-based alerts by email and Telegram, newsletters that a person reviews before they are sent, a trends dashboard and collections. Coming soon: reports in PDF, Word and PowerPoint, a public API, predictive alerts and dedicated connectors for regulation, tenders and patents.

Which sources can I connect?

RSS and Atom feeds, sitemaps, pages whose changes you want to follow, newsletters that reach a mailbox, RIS, CSV and BibTeX files, PDF, Word, Excel and PowerPoint documents, images with text recognition, and the open databases arXiv and OpenAlex.

What happens with repeated news?

Exact duplicates are not stored twice. Near-duplicates are grouped under a representative item, so you see a single entry with the number of repetitions.

Can I add my own documents?

Yes. You can upload PDF, Word, Excel and PowerPoint documents, and images with text recognition, as well as import references in RIS, CSV or BibTeX format.

How do I get started?

By creating your free account. When you sign up, an assistant suggests the source package for your sector and sets up your first channel and your first watchlist.