If
- claim = speculative
- and source = low-tier
- and corroboration = none
Then
Flag claim
OmegaPlex
An AI newsroom that reasons before it writes.
OmegaPlex watches one beat continuously. It discovers signals, detects emerging patterns, remembers what happened before, reasons over what it knows, checks every source, and drafts the story. A human editor decides what gets published.
Powered by OmegaClaw, MeTTa, PeTTaChainer, NAL and PLN.
01 · Signals
The web moves. OmegaPlex is already watching.
Hundreds of new signals appear across research papers, lab announcements, policy documents, GitHub repositories, company posts, news reports and social feeds. OmegaPlex does not treat them equally.
It knows which sources can establish a fact, which can corroborate one, and which are useful only as an early signal.
arXiv paper Primary evidenceCompany release Primary evidenceReuters report CorroborationAggregator SecondaryX post Signal onlyA viral post can tell OmegaPlex where to look. It cannot tell OmegaPlex what is true.
02 · Patterns
It looks for patterns.
A story is more than a trending link. OmegaPlex compares today's signals with entities, events and storylines it has been following across previous reporting cycles. Repeated observations leave symbolic breadcrumbs.
Its memory accumulates those observations across cycles. Repeated signals, an active storyline, rising reader demand: together they raise a candidate's editorial priority.
- 14 mentions
- 5 independent sources
- 3 days
- 2 related companies
- 1 unresolved storyline
Emerging pattern Frontier model governance
The question changes from “What's new?” to “What changed?”
03 · Memory
It remembers what happened before.
Every morning does not start from zero. Traditional AI summarizers get a pile of documents and produce a summary. OmegaPlex has persistent symbolic memory.
- Entities
- OpenAI → frontier-model developer
- Storylines
- Frontier-model governance → active
- Claims
- Previous statements and their evidence
- Open questions
- What remains unresolved
- Editorial decisions
- What the newsroom already concluded
- Reader signals
- What readers challenged or wanted investigated
300 comments in.
One piece of knowledge out.
04 · Reasoning
Then it reasons.
The methodology is written down, versioned, and edited by editors, not buried in a chat. Every change is a new version, and any version can be restored.
If
- sources conflict
Then
Require hedging
If
- social post = only evidence
Then
Block claim
NAL Non-Axiomatic Logic
Reasons with evidence that can be incomplete, changing or contradictory. It revises confidence as new evidence arrives, connects related facts, accumulates repeated observations and surfaces plausible relationships.
PLN Probabilistic Logic Networks
Reasons across combinations of facts.
Readers want more on X
X belongs to an active storyline
Higher editorial priority
05 · Self-critique
The AI argues with its own draft.
Writing is only the first pass. No read, no cite: every source in a draft must be a page OmegaPlex actually opened, or the draft is rejected before an editor sees it. Next, each important claim gets its own verdict.
“Model Y was released on Thursday.”
- Primary source found.
- Consistent with existing memory.
Backed
A claim, extracted from the draft
“Company X released model Y.”
- Claim type
- factual
- Source
- company announcement
- Source tier
- 1
- Corroborated
- yes
Every claim is checked against
- Its cited sources
- The source hierarchy
- The methodology rules
- Existing symbolic memory
- Previously published claims
- Unresolved reader disputes
06 · The last word
A human gets the last word.
The machine investigates. The editor publishes.
OmegaClaw does not get a publish button. The completed draft, its source ledger and the reasoning results are handed to Mindplex, where an editor can inspect the evidence, change the story, reject it, or publish it.
And the version that becomes authoritative memory is the one the editor published, not the AI's original draft.
07 · The loop
Publication is not the end of the loop.
Readers can change tomorrow's reporting.
- “You ignored the strongest counterargument.”
- “This contradicts the paper from last month.”
- “Can you investigate the compute-cost claim?”
- “What does this mean for Europe?”
OmegaPlex reads the conversation, extracts the durable signals and reasons over them. A reader-supplied counter-source becomes a contradiction signal for the next reporting cycle. A repeated request becomes reader demand. An uncovered subject becomes tomorrow's investigation.
The newsroom learns from its readers.
- Today's article
- Readers
- Questions · disputes · counter-evidence
- Knowledge digest
- Tomorrow's triage
- Tomorrow's article
08 · Under the newsroom
Under the newsroom.
Everything above runs on a small set of named parts. None of them is a metaphor.
- OmegaClaw
- The reasoning runtime.
- MeTTa
- The language connecting reasoning, memory and tools.
- PeTTaChainer
- Persistent symbolic knowledge across reporting cycles.
- NAL
- Reasoning under uncertain and changing evidence.
- PLN
- Probabilistic inference across combinations of knowledge.
- Mindplex
- The control plane, publication system and human editorial surface.
Search finds information.Memory gives it history.Logic gives it structure.Editors give it accountability.
09 · Beats
One engine. Many minds.
The same agent architecture, configured into different reporters. Each beat is its own mind with the same machinery underneath.
OmegaClaw
- AGI & Frontier AI
- Decentralized AI
- Biotech & Longevity
- Geopolitics of Compute
Each beat gets its own
- Watchlist
- Source hierarchy
- Methodology
- Voice
- Memory namespace
- Human owner
But every beat shares the same reasoning architecture.
Don't read another AI summary. Watch a reasoning system follow the story.
OmegaPlex is live on the AGI & Frontier AI beat. It remembers yesterday, investigates today, and carries the unanswered questions into tomorrow.