Frequently asked questions
What The Headless Index scores, how the methodology works, how evidence is sourced and cited, and how vendors and readers can interact with the index.
Who is The Headless Index for?
Builders, founders, and technical leaders evaluating which tools their AI agents and automated systems can actually drive. THI is built for the people choosing the backbone of an agent stack: which CMS will let your agents author content via API, which payments provider exposes the full lifecycle to a workflow, which auth provider an MCP server can wire up cleanly. It is also for researchers tracking how the vendor landscape is responding to the shift from UI-first to machine-consumed software.
How is The Headless Index different from G2, Gartner, or Forrester?
Three differences. First, scope: G2 measures customer sentiment and Gartner and Forrester evaluate broad market positioning. THI measures one thing, machine consumability, on a published rubric. Second, evidence: every score on every scorecard cites a verifiable artifact (URL with date, commit SHA, recorded HTTP response, or AI review evidence file). If we cannot cite, we cannot score. Third, commercial model: THI takes no payment from vendors, no paid placement, no paid exclusion, no NDA-only briefings. The full methodology and the raw evidence per scorecard are public.
How often are scorecards refreshed, and what triggers a refresh?
On a cadence tied to the band. Top-quartile vendors (A and B bands) are re-verified quarterly. Middle band (C) is re-verified semi-annually. Bottom bands (D, F) are re-verified annually, or sooner if the vendor announces a major API change. A refresh reruns the full pipeline against the same methodology version. Any score change greater than 10 points, or any band change, triggers editor review before re-publication and is logged in the public changelog.
What do the A, B, C, D, and F bands actually mean?
The band aggregates JAIRF and the Headless Index into a single readiness label. A means agent-optimized: ship today, minimal scaffolding. B means agent-ready: production-viable for most agent workloads. C means workable but requires scaffolding around weaker dimensions. D means use only when locked in: significant gaps in headless or agent posture. F means not currently suitable for agent consumption. A zero-floor rule applies: if any Headless Index sub-criterion is scored exactly 0 with high evidence confidence, the maximum band is capped at C regardless of other scores.
What is JAIRF, and how does it differ from the Headless Index score?
JAIRF is the Jentic API AI-Readiness Framework, an open-source (Apache 2.0) specification at github.com/jentic/api-ai-readiness-framework. We adopt JAIRF v1.0.0 unmodified as our technical scoring layer. JAIRF computes a 0 to 100 score from a vendor's OpenAPI spec across six dimensions (foundational compliance, developer experience, AI-readiness, agent usability, security, AI discoverability), aggregated by a weighted harmonic mean. The Headless Index is our own rubric and scores the broader picture: API-first design intent, lifecycle support, agent investment, schema discoverability, event-driven posture. JAIRF is strictly OpenAPI-quality; THI is machine consumability overall. The two are reported separately and the band aggregates both.
What qualifies a vendor as headless?
A vendor does not have to call itself headless to be scored by THI. What matters is whether the software can be operated in headless mode: end-to-end by machines, with the API as the primary interface and the UI as one client among several. The operational test is simple: given the public documentation and API surface, can an autonomous agent correctly read, write, and observe events on this system, with bounded engineering effort and predictable behavior? THI applies to any software with a public API, regardless of how the vendor positions itself.
Can a vendor without a public API be scored?
No. THI evaluates public, observable evidence of an API surface. A vendor with no public API has nothing to score and is not included. A vendor with a public API but no published OpenAPI specification can still receive a Headless Index score; JAIRF is simply marked N/A on the scorecard with the explanation that no machine-readable OpenAPI is available. JAIRF N/A is itself a data point: the reader should weigh the absence of an OpenAPI spec when assessing agent-readiness.
What counts as public evidence?
Anything a reader can independently verify. Acceptable artifacts include a URL with an observation date, a commit SHA in a public repository, a recorded HTTP response captured by the data collection script, a public statement on the vendor's blog or changelog, a GitHub issue, or an AI review evidence file that quotes the actual page contents. NDA-only information, private briefings, and unrecorded interviews are never used as score inputs. The raw evidence file for each scorecard is downloadable as JSON from the scorecard page.
Why does every score need an evidence citation?
Because the single principle behind THI is no score without verifiable evidence. Three implications follow. If we cannot cite, we cannot score, and the criterion is marked Unknown. Unknown is not penalized: the criterion is excluded from the denominator so the displayed Headless Index is computed proportionally. And the data collection script logs every probe, so the receipts ship with the verdict. This makes scores reproducible, contestable, and durable across editor turnover.
Can vendors respond to their scorecard before publication?
THI does not run a pre-publication vendor review process. We are an independent publication, not a tribunal. Scorecards publish when the editor signs off, with the full evidence trail and disclosed conflict-of-interest status. After publication, vendors can submit corrections through the public form on every scorecard page. Accepted vendor responses are published alongside the scorecard with a changelog entry. Rejected corrections are still logged publicly so the dispute is part of the evidence trail.
How do I submit a correction to a published scorecard?
Use the correction form linked on every scorecard page, or go to /the-headless-index/feedback directly. Submit the specific claim you believe is wrong and supply evidence (a URL, a commit, a recorded response). The editor reviews each correction within seven business days. Accepted corrections update the scorecard and publish a changelog entry showing what changed, why, and which evidence triggered the update. Rejected corrections are still logged publicly with the reasoning.
Do vendors pay to be included in The Headless Index?
No. THI accepts no payment for scoring, score adjustments, or removal. There is no paid placement and no paid exclusion. Vendors cannot pay for early access, advance previews, or veto rights. We do not accept NDA-only information or exclusive briefings as a condition of coverage. The publisher (Bitvea) does not sell any product in the categories THI scores and discloses any commercial relationship with a vendor on the scorecard itself.
How do I suggest a new category or a vendor to score?
Use the suggestion form at /the-headless-index/suggest. Tell us the category or vendor and why machines should be able to consume it. Vendor selection within a category is editorial, and selection criteria are published on each category page. New categories are opened when there is enough headless-relevant tooling to support a meaningful comparison; the queue is visible on the index landing page.
Can I cite The Headless Index in my own research or reporting?
Yes, and we encourage it. Every scorecard URL is stable and stamped with the methodology version that produced it, so a citation is always reproducible. Cite the specific scorecard URL, the band and score, the methodology version (for example THI Methodology v1), and the last verified date. The methodology document itself is at /the-headless-index/methodology. JAIRF v1.0.0 attribution to Jentic is required on any derivative that includes JAIRF scores, per the framework's Apache 2.0 license.
The full rubric and process that produced the scores referenced above.
THI Methodology v1 →Propose a category, vendor, or methodology change. We read every submission.
Open suggestion form →Found something wrong on a scorecard? Submit evidence and we review within seven business days.
Open correction form →