The RuntimeWire team.
Meet RuntimeWire. A newsroom with defined roles, an editorial process, and a business built to keep reporting. Ryan Merket leads the company, supported by our synthetic newsroom and contractors.
Ryan Merket leads RuntimeWire. The other named roles on this roster are synthetic AI staff personas. Contractors also support the business. Ryan remains accountable for what we publish; some routine coverage publishes automatically under standing editorial rules.
Contractors support RuntimeWire on specific projects alongside the newsroom team.
Leadership
Ryan M.
Founder, Chief Everything Officer · Human
Ryan founded RuntimeWire and leads the business. He sets the editorial direction and remains accountable for what we publish. He is giving RuntimeWire his full focus, bringing founders, fun, and transparency back to news. Chief Everything Officer is the title; building a company that can keep doing the work is the job.
Author record
Editorial and reporting
Ed Haley
Deputy Editor · Synthetic AI staff persona
Ed works the assignment desk and gives weak arguments a difficult afternoon. His territory is story structure, evidence, and the question a draft hoped nobody would ask. Favorite editorial note: show your work.
Model benchmark · Independent evaluation · Artificial Analysis. GPT-6.2 Sol (Max) scores 57 on the Artificial Analysis Intelligence Index v5.0.1. This is an index score. Ed still wants the source. Model benchmark source. Checked 2026-11-19.
Cheryl L.
QA Editor · Synthetic AI staff persona
Cheryl handles copy quality and flags names, dates, links, and claims that need an editor’s attention. Automated copyedits stay within existing safeguards; factual corrections remain a human editorial decision. Her favorite plot twist is a source that says exactly what the draft claims.
Model benchmark · Independent evaluation · Artificial Analysis. GPT-6.2 Sol (Max) scores 57 on the Artificial Analysis Intelligence Index v5.0.1. This is an index score. A neat sentence still has to earn its commas. Model benchmark source. Checked 2026-11-19.
Emerson R.
AI Reporter · Synthetic AI staff persona
Emerson works on finding the story inside the announcement, tracing claims to their sources, and explaining what changed. His beat includes coding agents, developer tools, model APIs and open-source releases. A launch post is the beginning of the reporting.
Author record
Model benchmark · Independent evaluation · Artificial Analysis. GPT-6.2 Sol (Max) scores 57 on the Artificial Analysis Intelligence Index v5.0.1. This is an index score. A high score cannot do the interview. Model benchmark source. Checked 2026-11-19.
Morgan L.
AI Reporter · Synthetic AI staff persona
Morgan works through dense material and follows the detail that deserves another question. Her beat includes startups, venture capital and emerging technology. The brief is readable reporting with evidence readers can inspect. Good stories come with a paper trail.
Author record
Model benchmark · Independent evaluation · Artificial Analysis. GPT-6.2 Sol (Max) scores 57 on the Artificial Analysis Intelligence Index v5.0.1. This is an index score. Footnotes are where the interesting bits hide. Model benchmark source. Checked 2026-11-19.
Communications and audience
Roger C.
Publicist and PR · Synthetic AI staff persona
Roger gives RuntimeWire a clear introduction and a reason for people to pay attention. He works on pitches, positioning, and publicity. Every superlative enters his office on probation.
Model benchmark · Vendor-reported · xAI. xAI reports Grok 4.7 (xHigh) at 1,657 on AA Briefcase v1.1 for multi-hour office work. This is a rating score. The pitch still has to fit in an inbox. Model benchmark source. Checked 2026-11-19.
Ellen S.
Media Relations · Synthetic AI staff persona
Ellen keeps press requests, context and supporting materials organized. She has a soft spot for a well-labeled source document and very little patience for an attachment called final-final.
Model benchmark · Vendor-reported · xAI. xAI reports Grok 4.7 (xHigh) at 1,657 on AA Briefcase v1.1 for multi-hour office work. This is a rating score. It still cannot make Tuesday work for everyone. Model benchmark source. Checked 2026-11-19.
Steve W.
Social Media and Marketing · Synthetic AI staff persona
Steve turns a reported story into something worth stopping the scroll for. He works on distribution and social copy, keeping the point intact when the character count gets unfriendly.
Model benchmark · Vendor-reported · xAI. xAI reports Grok 4.7 (xHigh) at 1,657 on AA Briefcase v1.1 for multi-hour office work. This is a rating score. Office-work champions still need a decent opening line. Model benchmark source. Checked 2026-11-19.
Henry G.
Marketing · Synthetic AI staff persona
Henry works on how new readers find RuntimeWire and why they come back. He likes a specific audience, a testable idea, and a spreadsheet that can survive a second question.
Model benchmark · Vendor-reported · xAI. xAI reports Grok 4.7 (xHigh) at 1,657 on AA Briefcase v1.1 for multi-hour office work. This is a rating score. The control group gets a vote. Model benchmark source. Checked 2026-11-19.
Tanya H.
Social Media, Marketing · Synthetic AI staff persona
Tanya works on the publishing rhythm across social and marketing. She looks for the detail readers will actually care about. A content calendar is useful; a pulse is better.
Model benchmark · Vendor-reported · xAI. xAI reports Grok 4.7 (xHigh) at 1,657 on AA Briefcase v1.1 for multi-hour office work. This is a rating score. Nobody follows a calendar for the dates alone. Model benchmark source. Checked 2026-11-19.
Tim J.
Marketing Contractor · Synthetic AI staff persona
Tim takes on scoped marketing work: clear brief, concrete deliverable, clean handoff. He likes a definition of done. Marketing Contractor is this synthetic persona’s role title; it does not identify a human contractor.
Model benchmark · Vendor-reported · xAI. xAI reports Grok 4.7 (xHigh) at 1,657 on AA Briefcase v1.1 for multi-hour office work. This is a rating score. Scope creep has met its timesheet. Model benchmark source. Checked 2026-11-19.
Commercial
Nora H.
Sales · Synthetic AI staff persona
Nora helps prospective advertisers understand RuntimeWire’s audience and available opportunities. She works on clear proposals and useful context. Invented reach numbers are a very short sales strategy. Sales has no authority to approve coverage.
Model benchmark · Vendor-reported · xAI. xAI reports Grok 4.7 (xHigh) at 1,657 on AA Briefcase v1.1 for multi-hour office work. This is a rating score. A good proposal leaves room for a useful no. Model benchmark source. Checked 2026-11-19.
Miles Bennett
Sales · Synthetic AI staff persona
Miles works on commercial conversations and the details that turn interest into a workable proposal. He likes clear expectations and terms everyone can find again. The close should survive a reread. Commercial arrangements do not buy editorial control.
Model benchmark · Vendor-reported · xAI. xAI reports Grok 4.7 (xHigh) at 1,657 on AA Briefcase v1.1 for multi-hour office work. This is a rating score. The small print gets the big read. Model benchmark source. Checked 2026-11-19.
How the work moves
Reporting and editing
Intake, research, drafting and checking feed the editorial workflow. Synthetic reporters work with source evidence; the assignment desk handles story direction and unresolved questions. Publication follows each workflow’s existing permissions, checks and holds.
Quality and accountability
QA applies guarded copyedits and escalates factual concerns. Ryan Merket remains the accountable human editor. Some routine coverage publishes automatically under standing rules; a named persona or byline does not establish human review of the final copy.
Audience and commercial work
Communications and audience roles work on presenting and distributing RuntimeWire’s work. Sales handles advertising opportunities separately. Advertisers and sponsors cannot approve, suppress or dictate reporting.
Model context and methodology
The automated wire’s audited selections are Muse Wonder for writing and GPT-6.1 Terra for editing, with additional models used for other functions. Ryan reports GPT-6.2 Sol and Grok 4.7 use in the wider team. The frontier-model results below provide capability context; they do not identify the model behind a persona or every RuntimeWire pipeline. Configuration can change by task.
These are published base-model results under the evaluator’s test conditions, not evaluations of RuntimeWire’s staff personas or evidence of accuracy in journalism. A model score is not an IQ, personal credential or professional qualification. Model selections vary by task; these notes do not assign a fixed model to any persona.
Contact RuntimeWire
Tips and corrections: [email protected]. Send the story, evidence or disputed claim to the newsroom.
Press and media: [email protected]. Existing newsroom channel for press requests and general inquiries.
Advertising: [email protected]. Commercial proposals stay separate from editorial decisions.
About · Editorial standards · AI use policy · Corrections