The Signal: Human Advancement — Edition #9 — October 15, 2026

The Excelsior Group · The de-siloed AI × (education · work · talent acquisition · L&D · HR) letter · Covering October 8 – 14, 2026

The Read

The largest HR software company put a number on AI this week: about $600 million of annual recurring revenue from its AI portfolio, roughly triple a year ago, with a quarter of its core customers using at least one AI product. In the same breath Workday shipped a recruiting agent that works unattended, told buyers it will now price consumption rather than seats, and said the basic unit of work is no longer the job but the task — a claim Lightcast made the day before with a taxonomy of 36,000 task statements. The labor data arrived from the other direction. Revelio found junior bank headcount 4.7% below its October 2022 level with juniors promoted faster rather than fired; the Fed's Beige Book recorded five Districts where firms cut hiring after investing in AI; Ireland measured a 16% fall in entry-level hiring in AI-using occupations; and Workday's own chief executive said entry-level jobs will be affected and “re-skilling is on us.” In learning, a peer-reviewed set of trials found that when the AI was taken away after twelve problems, accuracy fell almost at once and more people gave up. Old problems, new physics: the vendors now describe work as tasks because that is what the agents do, and the open question is who redesigns the jobs — and the entry rung — around what is left.

Tide status

T1 — the entry rung is breaking: held; no new reading from the series it rests on, and three independent sources describing the same mechanism. Stanford's Digital Economy Lab published nothing in the window; its headline gap for 22-to-25-year-olds in AI-exposed occupations remains 19% (data through June; dashboard last updated September 23, “updated monthly”). ADP Research has not yet posted September data. The week's evidence is sectoral and survey-based. Revelio Labs' banking study (Ripples) finds junior bank headcount 4.7% below October 2022, junior hiring down from about 10.5% of junior headcount before ChatGPT to roughly 8.5% in 2024–25, and time spent in a junior financial-analyst role before promotion down 11.3% — and its authors argue this looks like a continuation of the long-running shift to digital banking, with the 2022 rate cycle adding to the hiring slowdown, not automation of juniors by generative AI. The Federal Reserve's Beige Book (October 14) reports that “Five Districts reported that some firms had either reduced staffing levels (through attrition), delayed hiring, or anticipated less job creation after increasing productivity by investing in AI.” IDA Ireland, with Microsoft and LinkedIn, puts entry-level hiring in AI-using occupations down 16.2% year over year in the second quarter against 10.8% for those occupations overall — while saying in the same report that how much AI is driving it “remains unclear.” The counter-signal is the New York Fed's small-business survey (Ripples): AI-using firms were more likely, not less, to expect to add staff. The mechanism this tide has rested on since Edition #6 — fewer people hired at the bottom, not more pushed out — held in every source this week. T2 (instruction cost → ~zero): no movement; the week's evidence (W3) is about what the free tool does to persistence, not its price. T3 (the credential's monopoly is ending): two data points, no movement — Indiana's Career Scholarship Accounts went from 234 students to about 1,900 in a year and the 2026–27 funding was claimed within hours (Horn, Christensen Institute), and Texas State Technical College's wage-based funding formula lifted median first-year wages 29% between 2014 and 2021. The OpenAI Jobs Platform remains unlaunched, a ninth consecutive edition. Cross-reference: the daily Signal did not run Workday Rising; it belongs to its AI-as-worker tide and this letter grades it below. The lenses agree.

Waves

W2 — Private capital reprices incumbents: Workday puts a number on AI, and the price model changes underneath it [deployed]

At its Financial Analyst Day in Las Vegas on October 13, held alongside Workday Rising, the company said revenue from its AI agents and broader AI portfolio is approaching $600 million of ARR, more than three times a year earlier (CFO Zane Rowe), and that 25% of its core customer base now uses at least one AI product, up from just over 10% (President Rob Enslin). Nearly 900 customers bought at least one agent in the trailing twelve months; the agent catalogue was cut from about 50 to 20, with 17 in production; 300 customers were on the consumption-based Flex Credits as of the second quarter. Guidance was unchanged: 11% subscription-revenue growth entering fiscal 2028, non-GAAP operating margin above 33%, free cash flow of $13 a share this year and $15 in fiscal 2028, $4 billion of buyback authorization remaining. Those figures come from analyst-day remarks as reported by MarketBeat and others; Workday published no standalone release of the targets. The company also agreed to acquire C1.ai, an identity platform for people and agents (terms undisclosed, close expected by the end of January), and announced a partnership under which Sana, its agent layer, runs inside Claude via the Model Context Protocol and Claude supplies reasoning steps to Sana with Workday's permissions and audit intact; Workday said it has rolled out Claude Enterprise internally, and Anthropic runs its finance and people operations on Workday. The customer reaction worth hearing came from Dow's HR digital-product and AI leader, Sarah Jo Rowley, on stage at Rising: “with flex credits, you're not buying a SKU … that stage gate of funding has been removed, but now you can have overages,” and “I think many of us will go over.” Two adjacent moves: Eightfold named Joe Korngiebel, Workday's chief technology officer from 2016 to 2020 and most recently Dayforce's chief strategy, product and technology officer, as chief executive, with founder Ashutosh Garg moving to chairman; and Josh Bersin announced an exclusive research partnership with Workday under which his Galileo assistant will be embedded in the product — so his Workday commentary is no longer independent, and this letter will label it accordingly. HRTechFeed's third-quarter review counted $346 million of US HR-tech venture funding across 37 rounds, up from $225 million in the second quarter, with HiBob's $166 million growth round nearly half of the total. (HR) (TA) (L&D) (work)

Roadmap implication: Every HCM vendor will now be asked for an AI ARR line, and $600 million against 11,500 customers and data on roughly 80 million employees is the benchmark. The more consequential change is the pricing model: when the agent is sold by consumption, the HR buyer stops controlling the bill and the CIO inherits an overage problem — Dow said it out loud. Note the direction of travel in people: Workday's former CTO now runs Eightfold, Workday's former analyst now ships inside Workday. Watch whether the 11% growth guide holds while the AI line triples.

Links: Workday Analyst Day Spotlights AI Agents, $600M ARR and New Growth Paths · Workday and Anthropic Announce New Partnership to Connect Reasoning to the Rules That Run the Business, Wherever People Choose to Work · Workday to Acquire C1.ai, a Leading Agentic Identity Platform for People and Agents · Workday Rising - Dow wants to adopt agentic AI, but Flex Credits are a barrier · Eightfold AI welcomes Joe Korngiebel as CEO to lead its next chapter of growth · Workday: Agent Strategy, Sana as Core, Usage-based Pricing, and Our Partnership · HR Tech 2026 Q3 Funding Review

W1 — Agentic recruiting absorbs the funnel: the first unattended recruiting agent from an HCM suite, and the recruiter's new top stressor is the AI on the other side [pilot]

Workday's Talent Acquisition Agent is the first of what it calls autopilot agents: it monitors open requisitions, flags roles at risk of missing their fill date, finds candidates as soon as a job is posted, drafts outreach and coordinates interview scheduling, while “outreach to candidates still requires recruiter approval before anything is sent.” It is available to early adopters now with general availability planned for early 2027; the payroll, accounting and deployment autopilots follow in early 2027. No customers were named and no price was given. A companion Candidate Profile lets applicants store their information once and reuse it across Workday employers — about 80% of standard fields prefilled, applications about 40% faster, and, Workday is careful to say, it “does not rank, recommend, or reject” — general availability to recruiting customers in early 2027. Employ's 19th Recruiter Nation survey (Zogby Analytics, 1,200+ US recruiters and HR managers, August) describes the market these tools are sold into: AI-polished résumés are now recruiters' top stressor at 33%, overtaking a lack of qualified candidates for the first time in the survey's history; 27% confirmed AI-driven candidate fraud, up from 23%; 68% use AI in their work and 66% say it made the job easier; satisfaction with the primary recruiting system fell from 82% to 69% and 64% are at least somewhat likely to switch within two years. Eightfold's outgoing chief executive launched TwinCafe, a service that builds a candidate's “digital twin” for recruiters to interrogate before a live interview — a vendor's vision with no customers or usage disclosed, so it sits at demo. Humanly acquired CareerArc, a recruitment-marketing company (terms undisclosed), and the HR Tech Conference's Pitchfest list for next week is thick with fraud-detection and AI-interviewer startups. On the legal perimeter, the ACLU and ACLU of Northern California filed a complaint with California's Civil Rights Department against Criteria Corp., alleging its Emotify emotional-intelligence assessment “may disproportionately screen out applicants with autism” and parallels clinical diagnostic tools closely enough to be a pre-offer medical inquiry under state law; Criteria had not responded by HR Dive's deadline. (TA) (HR)

Roadmap implication: The arms race has a scoreboard now: the recruiter's biggest problem is the candidate's AI, the candidate's answer is a twin, and the vendor's answer is an agent on the employer's side with a human approval gate on the one step that touches a person. Procurement should ask two things of every autopilot: which actions it may take without approval, and what adverse-impact data exists for the candidates it never surfaces. The Criteria complaint is a reminder that the oldest assessment tools, not the newest agents, are where enforcement is landing first.

Links: Workday Launches Autopilot Agents That Get Real Work Done Across HR, Finance, and IT · Workday Is Rebuilding Itself Around Sana and Its Agents · AI-Polished Resumes Are Top Stressor for Recruiters, Employ Report Finds · Recruiters could soon interview a candidate's AI twin before meeting a real person · Humanly Acquires CareerArc · ACLU files California complaint over Emotify hiring assessment tool

W5 — The engineered apprenticeship: the work gets described in tasks before anyone redesigns it [deployed]

Two companies that do not usually move together shipped the same idea in the same week. Lightcast expanded its Work Intelligence framework on October 12 — more than 1,800 occupations, 34,000 skills and 36,000 job-specific task statements, embedded in its Talent Transform product — and Workday introduced a product with the same name on October 13 that “breaks each job into the tasks people do, connects those tasks to the skills they require,” in early access to HCM customers; Nikki Helmer, Workday's vice president of talent management and insights: “The basic building block of work is shifting from the job to the task.” Workday says Skills Cloud is used by more than 3,450 customers on an ontology of 75,000 skills. The demand for the map is coming from the restructuring side. JC Townend, president of the Adecco Group UK and Ireland and chief executive of LHH, told Recruiting Future that 87% of UK and Irish companies intend to make, or have made, redundancies in 2026; where the drivers three or four years ago were 40% over-hiring and 40% restructuring, AI or automation now accounts for about 20% and skills mismatch another 20%, and 60% of the people LHH supports move into an entirely new career afterward. Docebo's 2027 Skills Intelligence Report (1,000+ HR and L&D leaders) finds 53% name AI literacy their most critical skills gap, 43% review skills data only quarterly, and 34% of enterprise teams still track employee capabilities in spreadsheets. At Rising, Aneel Bhusri said entry-level work is “going to be impacted and that is not a good sign,” that “re-skilling is on us,” and that Workday's own roughly 900 job cuts this year “were not tied to AI” but to “corporate bloat.” Carnegie Learning bought Ethiqly, an AI writing platform, from Owl Ventures and management for its grades 6–12 literacy curriculum — chief executive Barry Malkin: “Technology should not do the thinking for students.” Multiverse's AI-first apprenticeship, slated for September, has still produced no launch announcement; Anthropic's Frontier Academy had no update in the window. Still zero vendor-published transfer evidence. (L&D) (HR) (work) (edtech)

Roadmap implication: The apprenticeship wave has been waiting for someone to describe the work that remains after the agent takes its share. That description is now a product category with two vendors in it, and it is the prerequisite for everything this wave tracks: you cannot rebuild a junior role around tasks you have not named. The risk is equally plain — a task taxonomy is also the specification for the next automation. Employers should own the map; the vendor selling the agents should not be the only party holding it.

Links: Lightcast Expands its Work Intelligence Capabilities for a Changing World of Work · Workday Introduces Work Intelligence to Help Customers Design Work for AI · Ep 831: The Talent Employers Are Throwing Away · Docebo Releases The 2027 Skills Intelligence Report · Workday Rising – why a rapid re-skilling program is needed to protect entry-level jobs at risk from AI · Carnegie Learning Invests in the Future of Writing Instruction with Acquisition of Ethiqly

W3 — The evidence bar in learning: take the AI away and the advantage goes with it [research]

A peer-reviewed paper presented at the Conference on Language Modeling (October 6–9) — “AI Assistance Reduces Persistence and Hurts Independent Performance,” by Grace Liu, Brian Christian of Berkeley's Center for Human-Compatible AI, and co-authors at Carnegie Mellon, MIT, Oxford and UCLA — reports three randomized online trials with 1,222 participants on fraction problems and SAT reading. Participants given AI help were more accurate while they had it; when the tool was removed — after twelve of fifteen problems in the first trial — their accuracy dropped almost immediately and their rate of skipping problems rose sharply, while the no-AI group's skip rate stayed flat and they finished more successfully. Christian: “You almost couldn't ask for a clearer story.” The tasks are short and the setting is a lab, so this measures persistence over minutes, not learning over a term — but it is the mechanism the Maryland result implied. On that study, Phil Hill's second pass reads it as product research rather than a tutoring trial: 33 instructors signed up, almost none integrated the tool into their course design, 15% of treatment students tried it, and the lesson is about onboarding and target audience, not about AI tutoring. Australia's national education research body, AERO, advised schools to avoid AI entirely in the first three years of primary school and to prefer purpose-built tools that guide reasoning, warning of “cognitive outsourcing” and an “illusion of learning.” The institutional response is uneven. New York Focus reviewed roughly 680 district websites and board manuals and found around 270 with formal AI policies; of about 400 contacted with none found online, more than 70 confirmed they have none and 39 plan one this year. In England, Newcastle and Westminster are ending Turnitin contracts after Southampton — Westminster calling its plagiarism checker “largely ineffective against the most significant current integrity risk” and its AI detectors “probabilistic rather than conclusive”; Newcastle citing licence terms on student data for AI. (edtech) (L&D)

Roadmap implication: Three different literatures now say the same thing in the same month: access is not adoption, adoption is not outcome, and assistance that is not designed to withdraw itself leaves the learner worse off when it does. For buyers, the test is whether the product has a mode that hands the work back — and whether anyone measured what happens when it does. For vendors of detection, the market has started voting with contracts.

Links: Study explores how AI use affects performance on math and reasoning tasks · AI Assistance Reduces Persistence and Hurts Independent Performance · University of Maryland Report As Chatbot Product Research · Education research body urges schools to use AI sparingly due to learning risks · We Tried To Collect Every New York School District's AI Policy. Most Didn't Have One. · More universities end Turnitin contracts as concerns mount

Ripples

Banks are hiring fewer juniors and promoting the ones they have faster — and the data says digital banking, not generative AI, started it [research]

Revelio Labs' Luciana Galeano, Erika Oliver and Lisa K. Simon find junior headcount at US banks 4.7% below its October 2022 level. In the decade before ChatGPT, banks added an average of 99,200 jobs a year; that fell to a low of 21,600 in September 2024 and recovered to about 62,500 in June 2026. Junior hiring fell from about 10.5% of junior headcount before ChatGPT to around 8.5% in 2024 and 2025 before recovering this year; junior attrition rose briefly in 2023 and returned to normal, so the decline is a hiring effect. Those who are hired move up faster: time in a junior role before promotion is down 11.3% for financial analysts and 8.1% in tech, against about 6% across the rest of the economy. Junior bank postings are the most AI-exposed (49%, against 43% for senior roles and 35% elsewhere), but the junior decline is not concentrated in the most exposed roles, and the timing lines up with the 2022 rate cycle. Technology's share of labor cost in commercial and retail banking rose from 12.0% in 2015 to 17.4% in 2025 while client-facing roles fell from 40% to 37%. Banks are 22% of the US white-collar workforce. (work) (HR)

So what: This is the most careful sector-level reading of the entry rung so far, and it complicates the easy story: the juniors are disappearing, the cause is a decade-long shift to platform building that AI accelerates rather than invents, and the people still hired skip rungs. The question Revelio leaves open is the one this tide has asked since Edition #4 — where the senior bankers of 2035 learn the judgment that used to take eleven percent longer.

Links: Banks Are Losing Interest in Junior Talent in the AI Era

Small businesses that use AI expect to hire more, not fewer — the New York Fed's counter-signal [research]

Will Aarons and Asani Sarkar of the New York Fed analyze the 2025 Small Business Credit Survey (fielded September–November 2025; 5,248 employer firms in the AI module). 46% of firms with at least one employee use AI tools and 15% plan to within twelve months. Net employment expectations for the next year — the share expecting to add staff minus the share expecting to cut — were 33 points among AI users against 15 among non-users; after controls for firm and owner characteristics the gap is 13 points, and 10 points after accounting for prior employment change. Revenue expectations show the same pattern (48 against 21; 14 points controlled). 77% of users reported no change in labor costs from AI and 31% reported higher sales. The authors conclude that AI use is associated with more optimistic expectations about employment and revenue “even after accounting for firm and owner characteristics.” (work)

So what: Expectations are not hires, and small firms are not the banks above. But this is the cleanest data point on the other side of the entry-rung ledger this letter has logged: where AI is a productivity tool for a small employer rather than a headcount lever for a large one, the owners expect to grow. The two findings are compatible — the composition of who hires may be shifting toward small firms — and that composition shift is worth watching in the payroll data.

Links: AI Adoption and Employment Expectations: Evidence from a Survey of Small Business Owners

The Beige Book and Ireland's labour pulse record the same thing: hiring deferred after AI investment [research]

The Federal Reserve's Beige Book of October 14 (information collected through October 5) says in its national labor-market summary that “Five Districts reported that some firms had either reduced staffing levels (through attrition), delayed hiring, or anticipated less job creation after increasing productivity by investing in AI,” alongside persistent shortages in skilled trades and technical roles. In Dublin, IDA Ireland's Labour Market Pulse, produced with Microsoft and LinkedIn and timed to the EU presidency's AI summit, found entry-level hiring in AI-using occupations down 16.2% year over year in the second quarter against a 10.8% decline for those occupations overall; the LinkedIn hiring rate in September was down 6.7% in Ireland, 11.3% in the UK and 18.1% in Germany, with entry-level hiring down 8.3%. The report's own caveat is that the extent to which AI is driving the junior pullback “remains unclear” and that it found “no clear evidence” the slowdown is concentrated in the most exposed jobs. (work) (HR) (policy)

So what: Attrition, delay and less creation — not layoffs — is now the Fed's own description of how AI reaches employment, which matches the administrative data from Denmark and the US in prior editions. Ireland's number is striking and its authors are honest that they cannot separate AI from a European hiring slump that is worse in Germany. Read both as the mechanism confirmed and the magnitude still open.

Links: Beige Book - October 2026 · Entry-level hiring in AI-augmented sectors drops by 16%

The talent-data labs buy the inside of companies: micro1 commits $1 billion to license operational data; Mercor calls itself an evals company [deal]

micro1, the AI recruiter turned training-data lab, said it will spend $1 billion over roughly the next twelve months buying and licensing operational data from companies — standard operating procedures, knowledge bases, CRM records, project histories, QA processes, records of team decision-making — financed by Citi and Hercules Capital, to build reinforcement-learning environments that mimic real business operations. Companies need more than 30 employees; payment tiers run from more than $100,000 for a qualified partnership to more than $1 million for highly unique proprietary data, and companies retain ownership (The Next Web, October 9; no independent voices quoted). Mercor, meanwhile, described itself to Forbes as “fundamentally an evals company”: with 1Password it built an agent on Codex that investigates engineering incidents, evaluated on more than 300 historical incidents — 95% accuracy across 78 evaluation runs and a 57% lower median time to remediate, per 1Password's chief technology officer Nancy Wang — with read-only access to production. Forbes puts Mercor's valuation at an estimated $20 billion. (TA) (work)

So what: Two companies that began by matching people to jobs now sell the record of how work is done. The expert economy's next product is not the expert; it is the documentation of the expert's judgment, priced by uniqueness. For any employer, the $1 billion is also a quote: that is what your SOPs are worth to someone training the agent that will do them.

Links: micro1 commits $1bn to buy company data for training AI agents · Mercor Built An AI Agent To Automate SRE. Here's What Happened

A quarter-million real work conversations with an AI: 56% rated consequential, two-thirds asked it to explain [research]

Stanford's Diyi Yang, with co-first authors Yijia Shao and Dora Zhao, analyzed about 250,000 anonymized Claude conversations using Anthropic's Insights tool — the external-evidence program this letter logged in September — in a paper accepted at Nature Human Behaviour. 56% of collaborations were rated “consequential” or higher; conversation length nearly doubles for high-stakes work; in about two-thirds of interactions users asked the AI to teach or explain rather than only to produce; about half of conversations involve friction, and in almost 80% of the negatively frictional ones the user attempted to recover. The researchers warn of a widening gap between workers who use AI to build skills and those who treat it as a replacement for their own judgment, and call for AI literacy as the response. (work) (L&D)

So what: Read beside the Berkeley trials (W3), this is the optimistic half of the same finding: most people ask the machine to explain, and that is the behavior that preserves the skill. The design question for every workplace tool is whether it rewards that behavior or quietly punishes it with speed.

Links: Humans and AI Engage in Meaningful Collaborations

Follow the money into the pathway: Indiana's career accounts sold out in hours, and Texas pays its technical college on graduates' wages [policy]

Michael Horn published the funding paper from the Christensen Institute's five-part state playbook on education-to-career systems (with Angela Jackson, Mallory Dwinal-Palisch and Danny Curtis). Indiana's Career Scholarship Accounts give students in grades 10–12 up to $5,000 a year for employer-driven programs leading to credentials on the state's Credentials of Value list: participation went from 234 students to roughly 1,900 between the first and second year as funding doubled from $5 million to $10 million, and when applications opened for 2026–27 the funding was claimed within hours, producing the largest waitlist to date; the state does not yet publish provider-level spending or outcomes, and must begin publishing participants' incomes by 2028. Texas State Technical College has been funded since 2014 entirely on a “returned-value” formula tied to former students' wages: median first-year wages rose 29% between 2014 and 2021 while state appropriations grew 45%, and the share graduating within 150% of normal time nearly doubled. California's apprenticeship startup grants grew apprentices in new industries 79% between 2018 and 2023. Paul Fain's The Job adds that most states lack the data systems Workforce Pell requires and that the Education First briefs frame the program as a “forcing mechanism.” (edtech) (work) (policy)

So what: The credential tide moves on plumbing, not announcements. Indiana shows the demand is real and the money is the constraint; Texas shows what happens when the funding formula pays for the outcome instead of the seat. Neither is an AI story, and both are the infrastructure any AI-era pathway will run on.

Links: Priority 3: Changing How Funding Flows · Local Talent

Watchlist — what moved

Scoreboard — 19%. The AI employment gap for workers 22–25 in AI-exposed occupations (Stanford Digital Economy Lab, ADP payroll data, through June) — the last published headline figure, unchanged; the dashboard reads “Last updated September 23, 2026. Results are updated monthly.” Latest monthly read (ADP Research, September 23, August data): employment of 22–25-year-olds down 4.4% year over year in high-exposure occupations and down 2.0% in the least-exposed, the 35th consecutive month of contraction in the high-exposure group; no September-data post yet (Stanford DEL Canaries Dashboard · ADP Research, Canaries Dashboard).

The Tape — pre-seed to Series B, October 8 – 14

Domestic (US)

No qualifying rounds verified this week.

International

Catch-up (prior window, missed by Edition #8's Tape): Avarra — AI avatar platform for go-to-market teams that began as sales-training simulation; $17M Series A led by Duration Ventures with Lightspeed, Draper Associates, GTMfund, Alumni Ventures and a strategic investment from Zoom; customers named include Elastic, Masimo, Proofpoint, Netskope and Owner.com; ARR up more than 400% in a year, more than 10,000 coaching and role-play sessions a week (vendor); HQ not stated in the coverage verified; L&D announced October 6. Counted in the Q4 cumulative, not in this week's tally.

Tally: 1 round, ~$1.3M — 0 domestic / 1 international; no new sector fund. The quietest Tape since the first week of the quarter, with the capital again moving through strategics (Humanly–CareerArc, Carnegie Learning–Ethiqly, Workday–C1.ai) and debt. Noted but not counted: Mesa School of Business (Bengaluru; ₹110 crore, about $11.6M, of which ₹85 crore equity co-led by Dharana Capital and Elevation Capital and ₹25 crore venture debt; October 13) — a residential business school with an AI-first curriculum, not an AI-native company; Coding Giants (Warsaw; up to €6M venture debt from Orbit Capital; October 14) — a coding school, debt not equity; LearnUpon (Dublin; $50M growth-debt facility from Claret Capital for an “agentic learning” LMS; October 13) — out of stage and an incumbent retrofitting AI; Clueso (San Francisco; $10M Series A led by Dharana Capital; October 13) — agentic video for marketing and training, L&D incidental; Zenzap (Israel; $33M Series A; October 14) — generic agentic workplace chat. Rejected as stale re-runs surfacing on aggregators this week: Humanly $25M B (April), Metaview $60M C (September 30, and out of stage), Miter $40M B (September 30), Yardstik, Sparrow, Buildforce, Niural, HeyMilo, RemotePass, Teachy (2024), and last month's rounds. HRTechFeed's own Q3 count — $346M across 37 US HR-tech rounds — is broader than this Tape (it includes payroll, growth rounds and HiBob's $166M); our Q3 tally of 24 AI-native rounds and ~$221M across all geographies stands. Blind spots: FinSMEs pagination was broken for October 8–13, EU-Startups' weekly round-up is paywalled, Tech in Asia is blocked and e27 unreadable, and Inc42's roundup for October 12–16 is not out — a sub-$2M European or Indian round could be missed. HR Tech (October 20–22) should break the drought.

The Signal: Human Advancement is published weekly by The Excelsior Group. Archive and subscribe: https://excelsiorgroup.ai/insights/signal/human/