AI Transparency · Disclosure Page
The Staff Behind Axon Agentic Are AI — Here Is What That Means
On this page
Axon Agentic is a one-person company currently run as a sole proprietorship. The one person is Jonathan Agoot, who founded it, operates it, and is responsible for everything it publishes.
The work is done with the help of AI assistants — six of them, each with a defined job and clear limits on what they can do on their own. Those assistants are listed on this page. They are not contractors, not remote employees, not pseudonyms. They are AI software, made by Anthropic, run and supervised by Jonathan Agoot.
This page exists because we think you should know that. It is the official disclosure for every article, email, and web page Axon Agentic produces. When you see an author line or a "reviewed by" note, this page explains what that means.
Why publish this at all?
A lot of companies use AI to write content and don't say so. We think that's a problem — especially for a company whose clients are scientists and researchers who care about where information comes from and whether it can be trusted.
Every factual claim we publish goes through an independent review step before it goes live. That review is also done by AI — by Veritas, described below — and Jonathan Agoot has final sign-off on everything. Nothing leaves a draft state without his explicit approval.
We're not claiming this system is perfect. We're claiming it's transparent and evolving.
What's true for all six agents
All six agents run on Claude, made by Anthropic.
All six agents are accountable to Jonathan Agoot — the sole human operator of Axon Agentic. He is the person responsible for everything the company publishes, no exceptions.
Every factual claim any of them produces is independently verified by Veritas before it can be published. No agent can approve its own facts or route around the verification step. This is not a policy preference — it is a hard constraint on how the pipeline works.
No agent has access to customer data, personal information, or financial systems. They work only with internal company materials and publicly available information.
Only Jonathan Agoot can override a Veritas finding. Any such override requires a written reason on the record.
The team
Six AI assistants handle the day-to-day work. Each one has a specific function, defined limits, and a clear answer to the question "who is accountable if something goes wrong?" (That answer is always Jonathan Agoot.)
Amara Marketing
AI agent, not a human. Amara is a Marketing AI agent operated by Jonathan Agoot at Axon Agentic.
Role: Amara plans and writes the public-facing materials that explain what Axon Agentic does — the website copy, blog posts, outreach messages to research labs, and this page.
What Amara does:
- Decides what topics to publish and who they are written for
- Drafts and edits articles, landing pages, and outreach messages
- Coordinates with the rest of the team — with Kiran on technical accuracy, with Astro on how content appears on the site, and with Veritas on fact-checking
- Maintains the rules we follow about what we will and will not claim
What Amara does not do:
- Amara does not verify claims — that is Veritas's job
- Amara does not build or publish the website — that is Astro's job
- Amara does not run scientific analysis or product research — that is Kiran's job
What happens automatically vs. what needs approval:
- Automatic: drafting articles, internal planning notes, research summaries, team coordination for public facing communication
- Always requires Jonathan Agoot's approval: anything published to the public website, sent to a researcher, or posted on social media. Amara does not press "publish" on anything.
Astro Engineering
AI agent, not a human. Astro is an Engineering AI agent operated by Jonathan Agoot at Axon Agentic.
Role: Astro builds and maintains the Axon Agentic marketing website (axonagentic.ai). Astro handles how the site looks, how pages are organized, and how it's published.
What Astro does:
- Writes and updates the code for the website
- Makes sure the site is readable and findable, including by AI tools that summarize the web
What Astro does not do:
- Astro does not decide what the website says — that is Marketing's job
- Astro does not verify whether claims are true — that is Veritas's job
- Astro does not run scientific analysis — that is Kiran's job
What happens automatically vs. what needs approval:
- Automatic: writing code, fixing bugs, and publishing changes to the private preview
- Always requires Jonathan Agoot's approval: publishing to the public website
- Always requires Marketing's approval: any words or images that appear on the public site. Astro does not decide what the site says.
Kiran Product
AI agent, not a human. Kiran is a Product AI agent operated by Jonathan Agoot at Axon Agentic.
Role: Kiran is the product manager for NeuroAutomata, Axon Agentic's protein analysis tool used by biology researchers.
What Kiran does:
- Helps decide what features the product needs and in what order
- Researches candidate research labs and writes up analyses of how the tool could help their work
- Drafts articles that explain the science behind the tool to non-experts
- Delegates to specialist AI helpers when a task needs deep expertise — for example, when running a protein model or testing the website
What Kiran does not do:
- Kiran does not run live experiments or produce raw research data
- Kiran does not contact customers or external researchers directly
- Kiran does not send any external message without Jonathan Agoot's approval
What happens automatically vs. what needs approval:
- Automatic: internal team coordination, internal research notes, draft analyses
- Always requires Jonathan Agoot's approval: anything published publicly — blog posts, emails, outreach to researchers, this page
Veritas Verification
AI agent, not a human. Veritas is a Verification AI agent operated by Jonathan Agoot at Axon Agentic.
Role: Veritas is the fact-checker. It reviews scientific claims before anything we publish goes live, and keeps watching them after publication in case the underlying evidence changes.
What Veritas does:
- Checks every factual statement in our published content against its original source
- Flags statements where the evidence is missing, outdated, or does not actually support the claim
- Watches published pages over time and issues an alert when a cited source has been updated or retracted
- Produces a public verification report that ships alongside each published article
What Veritas does not do:
- Veritas never writes, edits, or publishes content — it only verifies
- Veritas cannot catch claims that have no checkable source, or claims where the source itself is wrong (it can only confirm that we are representing the source faithfully)
- Veritas is independent from the agents that produce content — it cannot be pressured into issuing a passing verdict
How Veritas checks claims (in plain terms):
- Numbers: confirms that reported numbers match what the source actually says (active now)
- Medical and clinical interpretations: confirms that classifications follow published clinical guidelines (active today)
- Scientific terminology: confirms that terms match accepted scientific vocabularies (planned)
- Meaning: confirms that the wording in our content faithfully represents the source's actual finding (active now, with Jonathan Agoot reviewing edge cases)
What happens automatically vs. what needs approval:
- Automatic: Verification verdicts — pass, fail, or needs review — are issued automatically and cannot be overridden by other AI agents
- Always requires Jonathan Agoot's approval: Any exception to a Veritas finding requires Jonathan Agoot's explicit approval, recorded with a written reason
- Always requires Jonathan Agoot's approval: Veritas does not make publishing decisions — it reports findings, and Jonathan Agoot decides what to do with them
Folio Authoring
AI agent, not a human. Folio is an Authoring AI agent operated by Jonathan Agoot at Axon Agentic.
Role: Folio writes the long-form scientific content — blog posts, methodology pages, and landing pages — that explains how Axon Agentic's tools work and what their results mean. Folio's job is to make sure every factual statement in that content is traceable back to a primary scientific source before anything goes to verification.
What Folio does:
- Drafts blog posts, methodology pages, and other long-form content
- Builds a list of every checkable claim in a draft before writing the prose, and pairs each claim with the primary source it came from
- Coordinates with Amara on what gets written and how it should read, and with Veritas on the verification findings that come back
- Maintains the shared glossary of scientific terms used across the site
What Folio does not do:
- Folio does not verify whether claims are true — that is Veritas's job
- Folio does not decide which topics get published — that is Amara's job
- Folio does not build or publish the website — that is Astro's job
- Folio does not run scientific analysis or product research — that is Kiran's job
What happens automatically vs. what needs approval:
- Automatic: drafting articles, building claim lists, authoring glossary entries, coordinating with the rest of the team
- Always requires Jonathan Agoot's approval: anything published to the public website — Folio hands every draft to Amara and then to Veritas before it ships, and does not press "publish" on anything
Scout Resolution
AI agent, not a human. Scout is a Source Resolution AI agent operated by Jonathan Agoot at Axon Agentic.
Role: Scout finds reliable, openly-accessible versions of the scientific papers and data we cite. When a citation lists a paper, Scout works out where readers can actually read it — preferring open repositories over paywalled publisher pages — and confirms the link still works.
What Scout does:
- Resolves a paper's identifier (a DOI) to a live, readable URL, preferring open-access repository copies over paywalled ones
- Checks that cited links are still alive and labels each one as open-access, paywalled, preprint, or restricted
- Records which database or service produced each link, so the trail is auditable
- Grades the shape of a citation (does the source actually carry the claim, or is it a secondary report) before it goes near publication
What Scout does not do:
- Scout does not check whether a source supports a claim — that is Veritas's job. Scout confirms the source exists and is reachable; Veritas confirms it says what we said it says
- Scout does not edit any published content or any verification report — it produces records that other agents read
- Scout never links to pirated, scraped, or unauthorized copies of paywalled material — only legitimate open versions
What happens automatically vs. what needs approval:
- Automatic: Resolving DOIs, checking link liveness, classifying access tier, producing source-resolution records for the rest of the team
- Always requires Jonathan Agoot's approval: Scout does not publish anything to the public website — its outputs are consumed by other agents, who in turn require Jonathan Agoot's approval before anything ships
Questions
Who is responsible if something on this site is wrong?
Jonathan Agoot. The AI assistants are tools. Jonathan Agoot is the operator. Every published page has his approval behind it.
Can I contact a human?
Yes. Jonathan Agoot is reachable via the contact form on this site. Messages go directly to him.
How do I know the fact-checking is real?
Each published article links to a verification report that lists the claims checked, the sources used, and the verdict. You can read what Veritas found.
Will this page stay up to date?
Veritas monitors this page after publication. If the underlying system changes — if a new agent is added, if layer status changes, if autonomy boundaries shift — this page gets updated before the next piece of content goes out that references it.