Method
How we research
Acentric Research studies where emerging technologies intersect: which organizations are acting on those intersections, and how each is likely to develop over three years or more. Every document follows the same written method, so reports can be compared with each other and with later updates.
What we cover
We cover six technologies: AI, biotechnology, energy, materials, quantum, and space technology, and the places where two or more of them cross.
- An Intersection is a pair of technologies, such as AI × Biotechnology. Six technologies give 15 Intersections.
- A Junction is a specific focus within an Intersection, such as drug discovery within AI × Biotechnology. An Intersection may hold several Junctions.
What counts as an intersection. Each technology must make a material difference to the other: its design, operation, efficiency or efficacy. Using one technology as a general tool in another field is not enough. If removing one technology leaves the outcome largely unchanged, it is not an intersection. Creating demand for another technology is not enough either.
Who we cover. Public and private companies, large multinationals where the intersection could meaningfully affect their results, research institutions and government programs. A large company is included on public evidence, such as a dedicated unit, disclosed investment or named products, not on general statements of interest.
Time horizon. The central question is how each intersection develops over three years or more. Each investigation documents the current state of each technology, but near-term developments matter only for what they say about the long-run path.
How the research is produced
The research is produced by an AI research agent working under a written charter, with a human reviewer.
- The reviewer chooses the technologies and starting topics, approves the direction of each investigation, and approves every document before it is published. The research is guided, not autonomous.
- The agent works in recorded cycles. Where the answer is unknown, it writes down what it expects to find before running each step, then records what it found and what changed. Dead ends are recorded with the same care as findings.
- Rules before results. Before a Junction is studied in depth, its research plan fixes the rules that decide every count: who is included, cut-off dates, and how outcomes are classified. A rule changed after the results are known is disclosed, and the affected numbers are reported both ways.
Sources
Every factual claim is footnoted to a public source that was actually read. Each web link carries the date it was accessed.
- Primary records first: SEC filings, patent records, ClinicalTrials.gov, FDA databases, and government grant and contract records.
- Research literature: peer-reviewed journals and preprints (labeled as not yet peer-reviewed).
- Respected news organizations and trade publications, for events and reported facts, alongside the primary record where one exists.
- Published work of established analysts and institutions, such as the IEA, national laboratories and university research centers.
- Company materials, such as investor presentations and press releases, always labeled as company claims.
We use public, freely accessible data only: no paywalled content we cannot read in full, no logins, and no non-public information. Social media, forums and anonymous sources may suggest leads but never support a claim.
Verification
Before a document is published:
- Every number is recomputed by script from the underlying data.
- A separate AI reviewer, which did not draft the document, checks every footnote against the claim it supports, and every link is re-checked.
- A second separate AI reviewer makes the strongest case that the document is wrong, overstated or one-sided. Each point is answered by revising the text or recorded as a limitation.
- The whole document is checked against our rules on investment language and statements about people.
- The human reviewer reviews the final text and approves publication.
The code and data behind computed numbers are published with each document as a frozen data appendix.
Standard measures
- Maturity uses the International Energy Agency's Technology Readiness Level scale (1 = initial idea, 9 = commercial operation, 11 = proven stable growth). Biotechnology also states regulatory stage: preclinical, Phase 1–3, or approved; for devices, cleared or approved.
- Activity is drawn from public records only: patents, research publications, government grants and contracts, and private funding rounds disclosed in SEC Form D filings. Each count states its source, query and date range.
- Signposts are dated, observable events that would show progress or failure in a technology, such as a trial readout or a first commercial plant. Each carries its evidence and a stated confidence. Signposts track the technology, never company or stock performance.
People
Executives, scientists and directors are described only through public-record facts: roles and tenure, prior companies and their documented outcomes, publications, patents, grants, and compensation and insider transactions as filed. We make no judgments of competence or character and do not speculate about motives.
Research, not recommendations
We analyze companies as businesses and technology developers, never as securities. A document may say which companies hold distinctive opportunities or face distinctive barriers, with the evidence and criteria stated. It never gives price targets, valuations, buy, sell or hold views, portfolio suggestions or predictions of stock performance. Stock moves and news cycles never trigger a publication.
What we publish
| Document | What it is | Length |
|---|---|---|
| Landscape Report | A scored scan of candidate Junctions across all 15 Intersections: method, scores and evidence, one section per Intersection. Refreshed periodically. | 4,000–6,000 words |
| Intersection Brief | A plain-language overview of one Intersection, introducing its Junctions. An initial scan, not a full investigation. | About 1,000 words |
| Junction Brief | A plain-language overview of one Junction: current state, opportunities, challenges and the sources a deeper study would use. Superseded, not deleted, when a Junction Report on it is published. | About 1,000 words |
| Research Note | One self-contained, verified finding, published before the full report is ready. Provisional until its Junction Report is published. | 800–2,500 words |
| Junction Report | Our main deliverable: a full investigation of one Junction, covering its ecosystem of organizations, people, signposts, barriers and open questions. | 3,000–8,000 words |
| Report Update | A dated revision of a published report, made when a signpost occurs, new evidence contradicts a published claim, or a report is 12 months old. | As needed |
Every report and brief opens with a plain-language Overview and carries a dated changelog. A finding that an intersection is not real, or cannot be assessed from public data, is published like any other.
Corrections
Published errors are listed with the date found and the fix.