The OS powering Assistantly
All issues

Intelligence Brief

Week of
November 2, 2026

Your weekly edge, the research, frameworks, and signal that separate operators from spectators. Read and summarized so you do not have to.

Published Monday, November 2 at 6:00 AM PT. 12 sources, 8 minute read.

01

The Human Intelligence Edge

What makes humans irreplaceable in the AI era.

Supply Chain ForumOctober 16, 2026

Generative artificial intelligence and managerial cognition in turbulent supply chains: a real-time empirical analysis

Augustine Okeke and Ifeanyi Ugbebor

Why it matters. 200 senior supply-chain managers in a randomized scenario experiment, where GenAI support tracked with higher cognitive agility, decision accuracy, adaptability and faster response. Scenario work is not a quarter of real disruption and self-rated agility carries the usual inflation, so hold the size of the effect loosely. The conditional is what survives. The uplift ran through verification, domain expertise and governance, which means it showed up for managers who could check the output and worked inside a process that made them. The artifact at the end looks the same either way. Hire the person who can tell when the model is wrong about their own field, because nothing downstream of the decision will tell you whether anyone did.

Read the source
PNASMay 6, 2026

Robust associations of emotional intelligence with human flourishing: A second-order meta-analysis

Tyler J. Robinson and Ethan Zell

Why it matters. 62 meta-analyses, more than 3,000 studies, over a million participants, and emotional intelligence held a positive association with flourishing including workplace outcomes. At that scale this is about as close to settled as the literature gets, with two asterisks the authors raise themselves: heterogeneity is high, and the effects run strongest for self-report measures, which is exactly where the construct is weakest. Read it as license to treat socio-emotional skill as a real asset, not as license to buy a questionnaire. The measurement caveat is the practical instruction. What a person did in a hard conversation is collectable in a reference call; what they think of their own empathy is not evidence of anything.

Read the source
APA Monitor on PsychologyJuly 1, 2026

How AI is reshaping human skills and thinking

Zara Abrams

Why it matters. A survey of 319 knowledge workers found greater confidence in AI tracked with less reported critical thinking, and a Polish natural experiment saw unaided polyp detection fall 6 percentage points after AI entered the workflow. The survey is self-reported and the endoscopy result is observational, so neither one establishes cause. The endoscopy number is still the one to sit with, because it measured what trained people could do with the tool switched off, in a domain where the outcome is a tumor found or missed. Assisted output is evidence that a workflow ran. If you want evidence of judgment, the process has to contain a moment where the assistance was not there.

Read the source
02

The Broken System

Why resumes, interviews, and hiring infrastructure are failing.

Harvard Business School and AccentureSeptember 2021

Hidden Workers: Untapped Talent

Joseph B. Fuller, Manjari Raman, Eva Sage-Gavin, and Kristen Hines

Why it matters. 2,275 senior leaders and 8,720 workers across the US, UK and Germany: among employers running recruiting management systems, more than 90% used them to filter or rank, 88% said qualified high-skill applicants were vetted out for not matching exact job-description criteria, and 94% said the same of middle-skill applicants. The data are from 2020 and the stack has turned over twice since, so this is a baseline rather than an audit of your funnel. The mechanism did not change with the vendors. A filter tuned to exact criteria manufactures false negatives in volume, and the people it drops never appear in any report anyone reads. The only way to learn your own rate is to pull a sample of rejections and read them.

Read the source
ACM EAAMO '25November 4, 2025

AI Self-preferencing in Algorithmic Hiring: Empirical Evidence and Insights

Jiannan Xu, Gujie Li, and Jane Yi Jiang

Why it matters. 2,245 human-written resumes, 24 occupations, simulated screening pipelines, and LLM evaluators preferred the versions their own model had rewritten even with content quality controlled: same-model candidates were 23% to 60% more likely to be shortlisted. Simulated pipelines are not your pipeline and the sizes will move as models turn over. The direction is the problem. A candidate whose tool happens to match your screener's model collects an advantage that has nothing to do with the work, and neither side can see it happening. It is a direct argument for scoring something the model did not write: a work sample, a live problem, a reference who watched the person do the job.

Read the source
DeloitteMay 5, 2026

2026 Talent Acquisition Technology Trends: The new imperative

Bhawna Bist

Why it matters. More than 9,000 business and HR leaders, 95% of executives worried about the accuracy of candidate skills data, and 5% of organizations reporting significant progress on workforce data quality. A survey of executives measures their confidence rather than their data, so treat the pair as the shape of a hole rather than a measurement of one. The hole is real enough. Resumes, credentials, portfolios and interview answers all became cheap to generate in the same eighteen months, which retired every proxy at once. Polish in your pipeline is now a signal about tooling, not capability. What replaces it is authenticity checking and job-relevant evidence, and that costs time per candidate rather than another seat license.

Read the source
03

The SupaHuman Standard

The behavioral science of who actually thrives.

European Journal of EducationOctober 27, 2026

Workplace Adaptability and Development in Hybrid Environments: The Role of Cultural Intelligence, E-Leadership and Psychological Well-Being

Gaurav Kumar, Ka Yin Chau, Ankita Manohar Walawalkar, Vimal Kumar, and Massoud Moslehpour

Why it matters. 303 full-time hybrid workers in India, with cultural intelligence, e-leadership and psychological well-being each positively associated with perceived job performance. Cross-sectional, self-reported and snowball-sampled: three reasons to read it as directional and none to throw it out. Distributed teams run on context-reading nobody writes down, catching that a message landed wrong, knowing which thread needs a call instead of a reply. Those are behaviors, visible inside a paid trial week and absent from every resume. Note also what the outcome variable is. Perceived performance is a rating, so pair whatever you observe with something the person actually shipped on a deadline.

Read the source
Frontiers in PsychologyMay 8, 2026

How leadership emotional intelligence promotes team innovation: parallel mediating roles of psychological safety and knowledge sharing

Suxuan Xing, Xiaomei Yu, Jun Yu, and Jiajia Chen

Why it matters. 65 leaders and 391 team members across 65 teams, multi-source: leadership emotional intelligence related to team innovation, with psychological safety carrying 39.8% of the total effect against 6.5% running through knowledge sharing. The split is the finding. Safety did six times the work that information-sharing did, which puts the lever on whether people will contradict the lead rather than on whether the documentation is current. Correlational design, one national context, so causation stays open. For hiring a manager it still changes the question. Stop asking how they communicate and ask what happened the last time somebody on their team told them they were wrong.

Read the source
Journal of Applied Psychology2025

How and for whom using generative AI affects creativity: A field experiment

Shuhua Sun, Zhuyi Angelina Li, Maw-Der Foo, Jing Zhou, and Jackson G. Lu

Why it matters. 250 employees at a technology consultancy, randomized access to ChatGPT, and externally rated creativity rose most for the employees who already used strong metacognitive strategies. A field experiment with outside raters is sturdier than most of what gets cited on AI and creativity, inside one firm and one culture. The result repeats the pattern running through this issue: the tool multiplied what the person brought to it. Planning, monitoring and adjusting your own thinking is the thing being multiplied, and an hour is enough to test it. Ask how a candidate would approach the problem before they touch it, then what they would check afterwards. Tool access is not the variable you are selecting on.

Read the source
04

The Future of Work, Without the Clichés

Where AI and human collaboration is actually heading.

Nature Human BehaviourOctober 26, 2026

Generative AI and the temporary upskilling of knowledge workers

Emma Wiles et al.

Why it matters. 487 management consultants, randomized: ChatGPT lifted performance on data-science tasks outside their existing skill set, and no independent technical knowledge was left once the tool was taken away. One profession, one study window, so nobody should read a year of real work into it. The finding is clean enough to plan around, and it rhymes with the persistence and motivation results from the past few issues. Assisted output and acquired skill look identical in a finished deliverable and are not the same thing. If the plan is that people learn by using the tool, the learning has to be built in on purpose: unaided reps, explanations in their own words, some check on what remains when the window closes.

Read the source
arXiv preprintOctober 30, 2026

AutoAnalysis: An automated, self-updating meta-analysis of human-AI collaboration

Michelle Vaccaro, Abdullah Almaatouq, and Thomas W. Malone

Why it matters. 420 peer-reviewed studies and 1,955 effect sizes, preregistered and PRISMA-guided: human and AI pairs beat humans working alone at g = 0.53, while in the smaller set of studies that actually tested synergy the pairs came in below the better standalone performer at g = -0.35. It is a preprint, heterogeneity is high and the review pipeline is automated, so hold the exact numbers loosely. The shape is what changes how work gets divided. Pairing reliably beats the unaided human and frequently loses to whichever side was already stronger at that specific task. Synergy is not the default setting. It is the result of deciding, task by task, which parts the person keeps.

Read the source
McKinsey Global SurveyAugust 2026

The state of AI in 2026: On the road to ROI

Dan Tinkoff, Lieven Van der Veken, Michael Chui, and Tara Balakrishnan

Why it matters. 1,719 respondents across 97 countries: 80% of AI users reported productivity gains, half said it helped them make better decisions, and 37% said AI contributed positively to EBIT. That is self-reported speed set against a self-reported P&L line, from respondents who opted into an AI survey, so the gap is directional rather than an accounting of anything. It is also the gap most operators are currently standing in. Individual uplift does not add itself up into enterprise value, and the firms reporting otherwise were the ones that redesigned the workflow and kept people in the loop. The question is not how many of your team use it. It is what the process around them was rebuilt to do.

Read the source
Previous issueWeek of October 26, 2026 All issuesBrowse the archive