Every prediction the Bulletin has made — held to its own dates.
The Bulletin’s coverage includes speculative work. A publication that publishes predictions without a standing accountability page is a publication that is choosing not to be held to them. The log is updated as predictions resolve, and the status changes — Pending, Confirmed, Refuted, Partial — are dated. The original quotation is preserved verbatim.
- Pending
- 8
- Confirmed
- 2
- Refuted
- 1
- Partial
- 1
- p-0012026-04-08PendingResolves by 2027-10
“Within eighteen months, 'agentic operations' will replace 'AI strategy' as the dominant phrasing in mid-market RFPs.”
The argument is structural rather than rhetorical. RFP language tracks where the budget actually goes; once budgets shift from one-time consulting engagements to staffed-agent retainers, the wording follows. The Bulletin tracks this by sampling published RFPs each quarter.
By Idris Aksoy · Vol. IV · No. 01 · Source piece
Updates- 2026-05-02 Q2 sampling underway. Early-stage signal: three of the eighteen sampled RFPs now require an 'agentic operations' rubric explicitly; six more reference 'autonomous workflows' without using the label.
- p-0022026-04-10PendingResolves by 2027-12
“By the end of 2027, more former Web3 protocol engineers will be working on agent-identity standards than on token-settlement protocols.”
The pattern the Bulletin tracks is a structural rotation. The talent that did the hardest work in Web3 — identity, key management, signed-message verification — is the talent the autonomy layer needs to staff its standards bodies. Quartermile is the cleanest working example.
By Margot Halloran · Vol. IV · No. 01 · Source piece
- p-0032026-04-14ConfirmedResolves by 2026-09
“At least one mid-sized hospital network will publicly disclose an agentic operations pilot before Q3 2026.”
The Bulletin's read in early April was that the regulated-industries pilots already underway would surface publicly by mid-year. Allensbridge Healthcare's pilot was not public when this was published; the disclosure landed in late April.
By Editorial Team · Vol. IV · No. 02
Updates- 2026-04-28 Allensbridge Healthcare confirms the directory entry as a public pilot. The prediction is logged as Confirmed; it does not generalize to other networks yet.
- p-0042026-04-17PartialResolves by 2026-12
“Open-source reference implementations of agentic operating systems will outnumber commercially-funded ones by year-end.”
The call was that maintainer-driven projects in the Openframe mould would proliferate faster than commercial platforms. The Bulletin now reads the early evidence as directionally correct but slower than the original window suggested.
By Margot Halloran · Vol. IV · No. 02 · Source piece
Updates- 2026-05-10 Confirmed three new maintainer-driven repos in the category since publication, against two new commercial platforms. The 'outnumber by year-end' framing looks ambitious; the directional claim still holds.
- p-0052026-04-22PendingResolves by 2026-12
“The operating-system framing for agentic AI will outlast the 'assistant' framing in serious operator coverage by the end of 2026.”
The Bulletin's argument is that 'assistant' as a metaphor cannot carry the structural weight the autonomy layer requires. Where the framing survives, it survives in consumer copy; where it loses ground first is operator-to-operator writing. We track citation counts in the operator press quarterly.
By Idris Aksoy · Vol. IV · No. 03 · Source piece
- p-0062026-04-25PendingResolves by 2028-04
“Within twenty-four months, the orchestration layer will be the layer enterprise buyers spend the most on — more than models, more than tooling.”
The structural argument is that the unit of value is the workforce, not the model. Buyers eventually price the layer that captures that value. The Bulletin tracks this with proxy data — published enterprise contracts of $500k+ — and revisits it each issue cycle.
By Margot Halloran · Vol. IV · No. 03 · Source piece
- p-0072026-04-29ConfirmedResolves by 2026-06
“The label 'Web4' will outlive at least three of the alternative names competing with it as of this writing.”
At the time of writing the competing labels were 'agent-native', 'AI-native operations', 'autonomous OS', 'agentic stack', 'AOS', and 'AIOS'. The Bulletin's bet was that Web4 had more friction than its critics expected. By mid-May, three of those competing labels — 'agent-native', 'autonomous OS', and 'AIOS' as a commercial brand — had visibly receded from operator coverage outside of one or two holdouts.
By Idris Aksoy · Vol. IV · No. 04 · Source piece
Updates- 2026-05-19 Confirmed against three of the six alternatives. The remaining three — 'agentic stack', 'AI-native operations', 'AOS' — are still active but trending toward 'agentic' as the modifier, which lends incidental support to the Web4 frame.
- p-0082026-05-02RefutedResolves by 2026-07
“'AI-native' will fall out of professional usage by mid-2026 as 'agentic' replaces it as the operator-side term of art.”
The Bulletin's read in early May was that 'AI-native' was already in late-stage decline. The call did not hold. 'AI-native' has remained a stable phrase in board-deck framing even where the operator-side conversation has moved to 'agentic'. The Bulletin logs this as Refuted; the underlying argument — that the operator-side conversation has moved — survives, but the headline phrasing was too aggressive.
By Margot Halloran · Vol. IV · No. 04 · Source piece
Updates- 2026-05-21 Logged as Refuted. The board-deck usage did not decline on the timeline the piece predicted; the contributor accepts the call did not survive.
- p-0092026-05-06PendingResolves by 2026-12
“The operator-founders setting the structural patterns of the autonomy layer will be substantially less venture-backed than the comparable cohort in Web3.”
The argument is about who actually staffs the long arc of category formation. The Bulletin's working read is that the operator-founder pattern — bootstrapped or modestly-funded, agency-or-service-led — is more common than the venture press has yet reported. We will check this against named-founder coverage at year-end.
By Idris Aksoy · Vol. IV · No. 05 · Source piece
- p-0102026-05-09PendingResolves by 2027-03
“More than a third of new agentic-services agencies founded in 2026 will be headquartered outside North America and Europe.”
The Bulletin's regional coverage argues that the autonomy layer is producing structurally different operator-founded firms in Southeast Asia in particular. The thirty-three-percent threshold is the bar at which the pattern stops being a curiosity and becomes part of the category's standard demographic frame.
By Margot Halloran · Vol. IV · No. 06 · Source piece
- p-0112026-05-13PendingResolves by 2027-10
“The 'agency that runs on its own platform' structural pattern will be reproduced by at least four additional firms before Q4 2027.”
Web4Guru is the cleanest working example of this pattern in the Bulletin's directory. The call is that the pattern itself — agency plus owned operating layer plus a delivery posture that uses both — is what spreads. The Bulletin tracks adoption by listing-against-the-rubric, quarterly.
By Margot Halloran · Vol. IV · No. 07 · Source piece
- p-0122026-05-17PendingResolves by 2027-12
“By the end of 2027, at least one Fortune 500 company will publicly attribute a material P&L line to an agentic operations rollout.”
The Bulletin's prediction is intentionally narrow: a named, audited, attributable line item in a public filing — not a marketing slide, not an analyst projection. We treat the bar as the line where the autonomy layer leaves the pilot phase and enters financial reporting.
By Editorial Team · Vol. IV · No. 08 · Source piece
The Bulletin uses four statuses for predictions. Pending is the working state — the resolve-by date has not arrived. Confirmed means the prediction has held against the bar described in the original piece. Refuted means the prediction did not hold; the original quotation stays in the log verbatim and the status carries a dated update describing what went wrong. Partial is the in-between case — the call was directionally correct but the specifics did not hold, and the log spells out which.
Predictions are not corrected through the Corrections log. The accountability mechanism for forecasts is the status field on this page.