ISSUE 038 SHIPS THURSDAY

The signal
for agentic AI
in learning.

Autonoma/Intelligence is a research observatory for the enterprise — tracking, auditing, and publishing on the agents quietly reshaping how organizations teach, train, and develop their people.

Subscribe to the brief → Read Issue 038
Reports · shipped

142

Across 14 quarterly issues, 38 weekly briefs, 90+ field reports.

Sources · indexed

4.2K

Earnings calls, filings, primary interviews, public pilots.

CLO readers

1,840

From Fortune-500 L&D, talent, and people-operations.

Claims · audited

100%

Every published claim carries a trace. Every brief, a human signature.

§ 01 — What we publish

Research at three cadences.

Quarterly reports for the argument. Weekly briefs for the motion. Field reports when the story can't wait.

QUARTERLY · REPORT №14Q2 · MMXXVI

The teaching agent, at close range.

Inside six quiet pilots — a new division of labor between the coach, the content, and the machine. 60 pages, 47 sourced claims, two years of primary interviews.

60 PAGES · 47 CLAIMSREAD →
BRIEF · 038THIS WEEK

Six pilots, one procurement seam.

A Fortune-100 industrial quietly moved 60% of coaching spend to an agent in four weeks. The RFP simply stopped appearing.

6 MIN · 12 CITATIONSREAD →
FIELD · 2404.17.26

Workday ships role-adjacent tutoring.

Inside the decision to place a tutor at every 1:1 touchpoint — and the compliance review that almost killed it.

4 MIN · 8 CITATIONSREAD →
INTERVIEW04.11.26

A CLO rewrites her 2027 budget.

Catherine Wu on moving from program line items to agent seats — and why procurement got there before strategy did.

12 MIN · AUDIOLISTEN →
DATA · RELEASE04.04.26

The agent-seat price curve.

Benchmark pricing by buyer-size cohort, six vendors, rolling quarterly. Downloadable. Licensed for internal use.

CSV · DASHBOARDOPEN →
§ 02 — The product

Signal — the pipeline, for your corpus.

The research pipeline that powers Autonoma is available to enterprise buyers as a governed research instrument. Point it at your documents. Let your editors drive.

Build your own observatory.

Signal is a governed agentic pipeline — calibrated for corporate learning research, but configurable to any domain. Human reviewers stay in the loop. Every claim is cited. Nothing ships until an editor signs.

  • PLANSub-question decomposition — planner agent breaks your topic into tractable, auditable questions.
  • INGESTCorpus-aware retrieval across your documents, public sources, earnings calls, and filings.
  • SYNTHClaim extraction with contradiction surfacing — never a smooth consensus when there isn't one.
  • CITEEvery claim carries a trace — source, confidence, adversarial challenge, human reviewer.
  • REVIEWHuman-in-the-loop, always — your editors approve, revise, or reject before anything ships.
Request access → Read the technical paper
autonoma · signal · run ● live
$ autonoma run --topic "agent-coach adoption"
# planning · 12 agents · 1,284 sources
 
plan → 7 sub-questions · 3 adversarial
ingest → 1284/1284 ok
synth → 47 claims · 12 contradictions
cite → every claim · audit-ready
review → @v.okafor · pending
publish → signal/brief/0038
 
——
"The contradiction between vendor claims
and DoD field retention is the story."
›
01 · PLAN

Decomposition.

Planner agent breaks the topic into sub-questions, including adversarial ones. Editor-approved.

02 · INGEST

Retrieval.

Corpus-aware search across your documents and public sources. Every hit logged.

03 · SYNTH

Extraction.

Claims, confidence scores, and contradictions surfaced — never smoothed away.

04 · CITE

Trace.

Every claim tied to a source, a challenge, and a confidence. Fully auditable.

05 · REVIEW · HUMAN

Sign.

Your editor approves, revises, or rejects. No publication without a human signature.

§ 03 — Governance

The pipeline is agentic. The signal is not.

Autonomy is our method — not our pitch. Every brief, every report, every claim is audited by a human editor before it leaves the building.

We do not advertise the machine. We advertise the judgment.
— Editorial charter, § 1.0
01Corpus — every source logged, timestamped, and licensable to republish.LOGGED
02Synthesis — claims extracted with explicit confidence scores.SCORED
03Adversary — every claim challenged by an opposing agent before publication.CHALLENGED
04Citation — every number, quote, and attribution traceable in one click.TRACED
05Editor — a named human reviewer approves before anything ships.HUMAN
06Corrections — published, timestamped, never quietly edited.PUBLIC
§ 04 — For

Written for the people making the calls.

Three audiences, one posture: briefed, not pitched.

For · CLOs & L&D Leaders

The weekly brief you read before the 8am.

  • Market motion, compressed to six minutes.
  • Vendor claims, independently audited.
  • Budget implications, named by line item.
  • Case studies your peers haven't published yet.
For · Instructional Designers

The field notes from practitioners who ship.

  • Pattern libraries for agent-assisted curriculum.
  • Pedagogical rubrics for evaluating agents.
  • What's working, by format and by cohort.
  • A community of 1,800+ designers in the trenches.
For · EdTech & L&D Investors

The category map before the category names itself.

  • Quarterly market reports on agentic L&D.
  • Pricing curves, retention benchmarks, adoption signal.
  • Founder dispatches from the operator side.
  • Licensed data for thesis-grade memos.