Publishing systems are often described through size: the number of writers, editors, developers, subscribers, products or markets. Size can matter, but it is not the first question. The first question is whether the essential responsibilities are present and connected.
At ORG Times, two people—Lalit and Nidhika Devgan—operate within a wider layer of AI-enabled tools, reusable designs, structured publishing functions and founder approval gates. The result is not an automated newsroom. It is a small human organization with a larger technical reach.
Key points
- Completeness means closing the responsibility loop, not imitating the headcount of a conventional newsroom.
- AI can expand working capacity while humans retain editorial authority and accountability.
- Reusable components reduce repeated labor but do not eliminate verification.
- A small system earns the right to scale by producing evidence under real use.
What this observation currently supports
- Status
- Observed in controlled production use
- Observation
- Two people, supported by AI tools and reusable infrastructure, are operating the connected ORG Times publishing loop.
- Supported claim
- A small human-led team can perform the complete workflow from observation through publication and measured learning.
- Uncertainty
- The current evidence does not establish scaled newsroom throughput, sustained revenue, audience growth or transferability to another operator.
- Next evidence
- Track production time, corrections, source quality, audience response and operator effort across repeated publications.
Begin with the smallest complete loop
A publication is incomplete when it can create a page but cannot establish why the page should exist, who approved it, what evidence supports it or what happened afterward. The smallest complete system therefore begins before writing and continues after release.
For ORG Times, that loop is observe, qualify, shape, approve, publish, measure and learn. Each stage can be modest. None should be imaginary.
- Observation preserves the original signal.
- Qualification tests relevance, evidence, authority and boundaries.
- Shaping turns material into understandable editorial form.
- Approval keeps consequential judgment human.
- Measurement and learning prevent publication from becoming an endpoint.
What AI changes—and what it cannot own
AI changes the amount of material a small team can examine, organize, draft, compare and package. It can help identify missing fields, generate structured alternatives, maintain consistency and prepare multiple publishing components from one approved source.
It cannot inherit the publisher’s accountability. A model does not own the relationship with the reader, carry the legal identity of the company, obtain consent by assumption or decide that an attractive claim is sufficiently supported. NIST’s AI Risk Management Framework likewise emphasizes defined roles and responsibilities for human-AI configurations and oversight.
- Use AI to reduce mechanical distance.
- Keep the source and uncertainty visible.
- Require human approval at consequential gates.
- Never convert fluent output into assumed evidence.
Two humans, many coordinated functions
A two-person publisher still performs multiple roles: observation, research, writing, editing, design, technology, distribution, commercial review and measurement. The answer is not to pretend that these roles do not exist. It is to make their boundaries explicit and let reusable infrastructure carry the repeatable parts.
A shared data structure can support a page, search metadata, article schema, social copy and measurement without requiring five separate reinventions. The human work shifts toward context, judgment, correction and relationship.
Capacity is not credibility
More output is easy to confuse with progress. A small publisher can produce a large volume and still have weak evidence, unclear attribution or no learning. Credibility grows when the system can show why something was published, what remains uncertain and how a mistake would be corrected.
This is why ORG Times describes itself as human-led and AI-enabled. The phrase is not ornamental. It assigns authority.
Scale should be earned through repetition
The next test is not whether the system can generate more pages. It is whether repeated use improves speed without reducing care, whether readers can understand the distinctions among editorial, sponsored and research work, and whether another trained operator could follow the same process.
A complete small system becomes a credible foundation when it can be observed, measured, corrected and taught. Until then, its promise should remain proportionate to its evidence.