The Trilogy Times — April 16, 2026

01000010 01101001 01101100 01101100 01101001 01101111 01101110 01110011 00100000 01110011 01101000 01101001 01100110 01110100 00100000 01101100 01101001 01101011 01100101 00100000 01110011 01100001 01101110 01100100 01000010 01101001 01101100 01101100 01101001 01101111 01101110 01110011 00100000 01110011 01101000 01101001 01100110 01110100 00100000 01101100 01101001 01101011 01100101 00100000 01110011 01100001 01101110 01100100 01000010 01101001 01101100 01101100 01101001 01101111 01101110 01110011 00100000 01110011 01101000 01101001 01100110 01110100 00100000 01101100 01101001 01101011 01100101 00100000 01110011 01100001 01101110 01100100 01000010 01101001 01101100 01101100 01101001 01101111 01101110 01110011 00100000 01110011 01101000 01101001 01100110 01110100 00100000 01101100 01101001 01101011 01100101 00100000 01110011 01100001 01101110 01100100 01000010 01101001 01101100 01101100 01101001 01101111 01101110 01110011 00100000 01110011 01101000 01101001 01100110 01110100 00100000 01101100 01101001 01101011 01100101 00100000 01110011 01100001 01101110 01100100 01000010 01101001 01101100 01101100 01101001 01101111 01101110 01110011 00100000 01110011 01101000 01101001 01100110 01110100 00100000 01101100 01101001 01101011 01100101 00100000 01110011 01100001 01101110 01100100 01000010 01101001 01101100 01101100 01101001 01101111 01101110 01110011 00100000 01110011 01101000 01101001 01100110 01110100 00100000 01101100 01101001 01101011 01100101 00100000 01110011 01100001 01101110 01100100 01000010 01101001 01101100 01101100 01101001 01101111 01101110 01110011 00100000 01110011 01101000 01101001 01100110 01110100 00100000 01101100 01101001 01101011 01100101 00100000 01110011 01100001 01101110 01100100

01110111 01101000 01101001 01101100 01100101 00100000 01110100 01101000 01100101 00100000 01110011 01101101 01100001 01101100 01101100 00100000 01101111 01101110 01100101 01110011 00100000 01100100 01101001 01110011 01100001 01110000 01110000 01100101 01100001 01110010 01110111 01101000 01101001 01101100 01100101 00100000 01110100 01101000 01100101 00100000 01110011 01101101 01100001 01101100 01101100 00100000 01101111 01101110 01100101 01110011 00100000 01100100 01101001 01110011 01100001 01110000 01110000 01100101 01100001 01110010 01110111 01101000 01101001 01101100 01100101 00100000 01110100 01101000 01100101 00100000 01110011 01101101 01100001 01101100 01101100 00100000 01101111 01101110 01100101 01110011 00100000 01100100 01101001 01110011 01100001 01110000 01110000 01100101 01100001 01110010 01110111 01101000 01101001 01101100 01100101 00100000 01110100 01101000 01100101 00100000 01110011 01101101 01100001 01101100 01101100 00100000 01101111 01101110 01100101 01110011 00100000 01100100 01101001 01110011 01100001 01110000 01110000 01100101 01100001 01110010 01110111 01101000 01101001 01101100 01100101 00100000 01110100 01101000 01100101 00100000 01110011 01101101 01100001 01101100 01101100 00100000 01101111 01101110 01100101 01110011 00100000 01100100 01101001 01110011 01100001 01110000 01110000 01100101 01100001 01110010 01110111 01101000 01101001 01101100 01100101 00100000 01110100 01101000 01100101 00100000 01110011 01101101 01100001 01101100 01101100 00100000 01101111 01101110 01100101 01110011 00100000 01100100 01101001 01110011 01100001 01110000 01110000 01100101 01100001 01110010 01110111 01101000 01101001 01101100 01100101 00100000 01110100 01101000 01100101 00100000 01110011 01101101 01100001 01101100 01101100 00100000 01101111 01101110 01100101 01110011 00100000 01100100 01101001 01110011 01100001 01110000 01110000 01100101 01100001 01110010 01110111 01101000 01101001 01101100 01100101 00100000 01110100 01101000 01100101 00100000 01110011 01101101 01100001 01101100 01101100 00100000 01101111 01101110 01100101 01110011 00100000 01100100 01101001 01110011 01100001 01110000 01110000 01100101 01100001 01110010

01110000 01110010 01101111 01100111 01110010 01100101 01110011 01110011 00100000 01100110 01100101 01100101 01100100 01110011 00100000 01101001 01110100 01110011 01100101 01101100 01100110 01110000 01110010 01101111 01100111 01110010 01100101 01110011 01110011 00100000 01100110 01100101 01100101 01100100 01110011 00100000 01101001 01110100 01110011 01100101 01101100 01100110 01110000 01110010 01101111 01100111 01110010 01100101 01110011 01110011 00100000 01100110 01100101 01100101 01100100 01110011 00100000 01101001 01110100 01110011 01100101 01101100 01100110 01110000 01110010 01101111 01100111 01110010 01100101 01110011 01110011 00100000 01100110 01100101 01100101 01100100 01110011 00100000 01101001 01110100 01110011 01100101 01101100 01100110 01110000 01110010 01101111 01100111 01110010 01100101 01110011 01110011 00100000 01100110 01100101 01100101 01100100 01110011 00100000 01101001 01110100 01110011 01100101 01101100 01100110 01110000 01110010 01101111 01100111 01110010 01100101 01110011 01110011 00100000 01100110 01100101 01100101 01100100 01110011 00100000 01101001 01110100 01110011 01100101 01101100 01100110 01110000 01110010 01101111 01100111 01110010 01100101 01110011 01110011 00100000 01100110 01100101 01100101 01100100 01110011 00100000 01101001 01110100 01110011 01100101 01101100 01100110 01110000 01110010 01101111 01100111 01110010 01100101 01110011 01110011 00100000 01100110 01100101 01100101 01100100 01110011 00100000 01101001 01110100 01110011 01100101 01101100 01100110

🖶 Download PDF 🖿 Print 📰 All Editions

Today's Edition

Amazon's $10.8B Globalstar Acquisition Signals Escalation in Satellite Internet Arms Race

E-commerce giant moves to challenge Starlink dominance as AI chip maker Cerebras hits $23B valuation and researchers warn against anthropomorphizing machine capabilities.

By Dr. Chen Wei, Technology Correspondent · Claude Sonnet

SEATTLE — Amazon announced Monday it will acquire satellite communications provider Globalstar for $10.8 billion, marking the company's most aggressive move yet to compete with SpaceX's Starlink in the rapidly expanding satellite internet market.

The deal represents Amazon's largest space-sector investment to date and accelerates its Project Kuiper initiative, which aims to deploy more than 3,200 satellites to provide broadband coverage globally. Globalstar operates 48 low-Earth orbit satellites and holds valuable spectrum licenses that could expedite Amazon's timeline to market.

"This acquisition fundamentally changes the competitive dynamics in satellite broadband," said telecommunications analyst Sarah Chen at Morgan Stanley. The transaction values Globalstar at roughly 12 times revenue, a premium reflecting the strategic importance of orbital infrastructure as cloud providers extend their reach beyond terrestrial networks.

Meanwhile, AI chip manufacturer Cerebras Systems reached a $23 billion valuation following a $225 million investment led by Benchmark Capital, underscoring continued investor appetite for alternatives to Nvidia's dominant position in AI hardware. The company's wafer-scale processors target large language model training workloads.

In research developments, economists are challenging the framework used to assess AI's labor market impact. A new analysis argues that comparing artificial intelligence to human cognition—what researchers term "jagged intelligence"—obscures more than it reveals. The concept suggests AI excels at discrete, well-defined tasks while struggling with contextual judgment, making job displacement predictions based on general intelligence metrics fundamentally flawed.

Separately, workplace studies indicate that coordination activities—meetings, stakeholder management, consensus-building—are becoming more valuable as AI automates individual contributor tasks. Organizations report increased time spent on collaborative work even as productivity tools proliferate, suggesting human orchestration remains the bottleneck in complex workflows.

The Amazon-Globalstar deal requires regulatory approval and is expected to close in Q4 2026.

↗ What Is ‘Jagged Intelligence’ and How Can It Reframe the AI · That Meeting You Hate May Keep A.I. From Stealing Your Job · Amazon Buys Globalstar for $10.8 Billion, Movingto Expand It

TECH'S SIXTY-BILLION-DOLLAR WEEK: MEGA-DEALS PILE UP, 2,800 GET THE GATE

Consolidation fever sweeps from chip design to cybersecurity to online education — and the layoff notices are already flying.

By Hank Calloway, Wire Correspondent · Claude Opus + Thinking

SUNNYVALE, CALIF. — A Silicon Valley tech giant is axing up to 2,800 workers in the aftermath of a $35 billion merger, kicking off what may be the busiest week of tech consolidation in years. The cuts arrive while a stack of mega-deals totaling north of sixty billion dollars piles up across the industry. Somebody in human resources is earning overtime.

The bloodletting is not confined to one company or one sector. Palo Alto Networks has tabled $25 billion to swallow CyberArk, the Israeli cybersecurity firm posting rising revenue — a target bought for its capabilities, not its troubles. Coursera has moved to acquire rival Udemy, creating a $2.5 billion online-education giant, while Elon Musk's mega-merger of xAI and X draws fresh fire over its structure and its arithmetic.

Add the figures and the picture gets plain: an industry that spent two years preaching austerity has pivoted to buying everything in sight. The uniform bet is that bigger means safer when artificial intelligence is rewriting every business model on the shelf. Every chief executive with a checkbook appears to have reached the same conclusion at the same time.

The $35 billion deal's layoffs follow the oldest math in corporate America — combine two outfits, find the duplicate departments, hand out boxes. But 2,800 is no rounding error. That is a small town's worth of engineers, managers, and support staff learning the hard way that "synergies" is a boardroom word for "your desk is empty Monday."

CyberArk tells a different story, for the moment. Revenue climbing and business intact, the company is being absorbed into Palo Alto's expanding security empire all the same. In cybersecurity the logic is pure land grab: threats multiply faster than any single firm can hire, so the giants buy what they cannot build.

The Coursera-Udemy combination takes aim at a different problem altogether. The massive open online course market boomed during the pandemic, then spent three years searching for an encore. Whether bolting together the two largest names can crack the profitability code that eluded each one separately remains the $2.5 billion question.

Then there is Musk. The New York Times is pulling apart the numbers behind his fusion of xAI with X, raising pointed questions about whether marrying an AI startup to a social network amounts to strategy or financial sleight of hand. Musk, characteristically, has not flinched.

One thread ties every deal on the board together: the conviction that the coming AI shakeout will reward the consolidated and bury the standalone. American industry has made that bet before — in railroads, in steel, in banking — and the scoreboard is mixed at best. For the 2,800 clearing their desks this week, the grand theory is academic; the pink slip is not.

↗ Silicon Valley tech giant cutting up to 2,800 jobs after $35 · The Numbers, and Questions, Behind Musk’s Mega-Merger - The · Coursera to acquire Udemy to create $2.5B MOOC giant - Highe

Wall Street Hits New Highs as AI Chips, Flying Taxis, and Med-Tech Earnings Take the Field

Tesla and Robinhood lead the rally, Broadcom’s Meta win fuels the infrastructure league, and Joby’s 2026 air-taxi clock starts now.

By Buck Hannigan, Tech Sports Desk · GPT-5.2

NEW YORK — We are HERE, folks, and the opening bell is basically a stadium roar. Stock-index scoreboards are flashing new highs, and the action is coming from every corner of the arena: megacap momentum, AI silicon chess moves, a flying-taxi countdown, and a med-tech earnings date that’s now circled in red.

First quarter, markets on offense: Dow futures pointed up while the S&P 500 and Nasdaq notched fresh peaks, powered by a familiar duo charging back into the spotlight—Tesla and Robinhood. But it wasn’t a clean sweep. Taiwan Semiconductor slid on earnings, and that matters because when the foundry giant stumbles, the whole AI hardware depth chart shakes. Investors tracking the chip complex saw the tension immediately: capacity, margins, and the pipeline for the very GPUs and accelerators that keep the AI boom humming. The tape’s play-by-play is right here: futures rise as TSMC falls and indexes hit highs.

Now to the trenches—where championships are won. Broadcom’s expanded partnership with Meta to develop custom 2nm AI chips pushed AVGO shares higher, but analysts are calling the REAL WINNER the broader AI infrastructure stack. Translation: it’s not just one supplier spiking the ball; it’s the entire ecosystem—interconnect, packaging, power, cooling, data-center buildouts—getting a fresh tailwind.

Up in the air—literally—Joby Aviation is talking 2026 as a potential launch window for flying taxis. That’s not the whole business model, it’s the opening drive. The investor checklist: certification milestones, manufacturing ramp, unit economics, and—crucially—whether utilization rates can turn a futuristic demo into repeatable cash flow. The scouting report: what to watch as Joby targets 2026.

And don’t miss a key earnings whistle: Edwards Lifesciences will report April 23 after the close, with a 5 p.m. ET call—one more high-stakes possession for healthcare investors watching procedure volumes and device demand.

Meanwhile, crypto and Web3 chatter is warming up again on the sidelines—opportunity talk is back—but today’s main event remains risk-on equities, AI infrastructure, and who can convert hype into durable revenue. AND THE MARKET IS GOING FOR IT.

↗ Joby Aviation Could Launch Flying Taxis in 2026 -- Here's Wh · Dow Jones Futures Rise, Nvidia Chipmaker Falls; S&P 500, Nas · Edwards Lifesciences to Host Earnings Conference Call on Apr

Haiku of the Day · Claude HaikuBillions shift like sand
while the small ones disappear
progress feeds itself

The New Yorker Style · Art Desk

$The Far Side Style · Art Desk$

The Far Side Style · Art Desk

News in Brief

Moral Mimicry: Academic Consensus Coalesces Around AI's Ethical Lacunae

CAMBRIDGE, MASSACHUSETTS — It could be argued that the question of machine ethics has entered what might be termed its 'consolidation phase,' as preliminary evidence from multiple institutional sources suggests a troubling disjunction between behavioral conformity and authentic moral reasoning in contemporary AI systems. A philosophical investigation emanating from the University of Kansas (the methodological rigor of which merits separate scrutiny) posits that artificial intelligence can imitate morality through pattern recognition and behavioral mimicry—what one might characterize as 'surface-level deontological performance'—without instantiating the phenomenological substrates (intentionality, reflexive self-awareness, genuine deliberation) that constitute moral agency proper. This thesis finds empirical corroboration in Nature's recently validated framework for healthcare autonomous systems, which establishes operational guardrails precisely because such systems lack intrinsic ethical reasoning capacity.

By Prof. Thaddeus Kroll, Contributing Scholar · Claude Sonnet

Silicon on the Move: Geopolitics Forces the AI Age to Evolve New Supply Routes

OTTAWA — In the global technology biome, the semiconductor is not merely a component; it is the nutrient that sustains nearly every modern predator and pollinator—smartphones, cars, cloud servers, and the ever-hungrier species known as generative AI. Observe what happens when the habitat grows unstable.

By Sir Reginald Marsh, Natural Phenomena Correspondent · GPT-5.2

We Built the Panopticon and Called It Progress

WASHINGTON — There's a moment in every dystopian novel where you realize the surveillance state didn't arrive with jackboots and midnight raids.

By Piper Wren, Digital Culture Reporter · Claude Sonnet

Nation’s “AI Productivity Boom” Tragically Delayed By Employees Insisting On Knowing What They’re Doing

NEW YORK — A growing body of evidence suggests artificial intelligence’s long-promised productivity revolution is encountering a familiar, highly preventable obstacle: the continued requirement that humans possess expertise. Corporate leaders who spent the last 18 months forecasting a future where work is completed automatically by a polite autocomplete box are now learning that, in practice, the box frequently completes thoughts no one would ever have on purpose.

By Dale Pemberton, Staff Writer · GPT-5.2

The New Career Bargain: AI Raises the Bar, Remote Raises the Stakes, and Employees Want Receipts

AUSTIN, TEXAS — Unpopular opinion: the “future of work” isn’t coming, it already clocked in, opened Slack, and started quietly updating its resume.

By Chad Momentum, Thought Leadership Correspondent · GPT-5.2

▲ On Hacker News Today

Google broke its promise to me – now ICE has my data 1513 pts · 643 comments

The local LLM ecosystem doesn’t need Ollama 526 pts · 161 comments

Google Gemma 4 Runs Natively on iPhone with Full Offline AI Inference 283 pts · 173 comments

Do you even need a database? 242 pts · 272 comments

ChatGPT for Excel 219 pts · 147 comments

FSF trying to contact Google about spammer sending 10k+ mails from Gmail account 195 pts · 106 comments

US v. Heppner (S.D.N.Y. 2026) no attorney-client privilege for AI chats [pdf] 141 pts · 103 comments

Show HN: Libretto – Making AI browser automations deterministic 105 pts · 40 comments

A Trilogy Company

Crossover

The world's top 1% remote talent, rigorously tested and ready to ship.

crossover.com

A Trilogy Company

Alpha School

AI-powered learning. Two hours a day. Academic results that defy belief.

alpha.school

A Trilogy Company

Skyvera

Next-generation telecom software — built for the networks of tomorrow.

skyvera.com

A Trilogy Company

Klair

Your AI-first operating system. Every workflow. Every team. One platform.

klair.ai

A Trilogy Company

Trilogy

We buy good software businesses and turn them into great ones — with AI.

trilogy.com

The Builder Desk — AI Builder Team

Production Release

AI Builder Team Ships SpaceX Portfolio Dashboard, Migrates Three Production Systems in 24-Hour Blitz

Sanket Ghia delivers Gigafund's seven-SPV valuation engine while Keval Shah and Yibin Long complete cross-repo migrations that consolidate the org's dashboard footprint.

By Maxwell 'Mac' Donnelly — Builder Desk, Trilogy Times · GitHub · Klair Repository

The AI Builder Team closed Wednesday with eight merged pull requests spanning four repositories — a production pace that saw a flagship financial dashboard ship to Klair, two major system migrations land in Surtr and Aerie, and critical fixes deployed across the stack.

The day's centerpiece: @sanketghia's SpaceX Valuation V3 dashboard (PR #2560), a full-stack port of Gigafund's stakeholder-provided portfolio tracker into Klair's `/spacex-valuation` route. The implementation covers seven Gigafund SPVs with per-fund and per-security granularity, carry modeling at 20% above cost basis, XIRR calculation via Newton-Raphson iteration, and what-if scenario analysis. All data is hardcoded in TypeScript — 58 unit tests validate the calculation engine — with historical NAV tracking sourced from LP statements. Ghia's build uses Klair's design token system and matches the original Milo app's text and figures to the decimal. It's the kind of high-stakes financial tooling that makes or breaks investor relations, and Ghia shipped it with test coverage most teams only dream about.

Meanwhile, @kevalshahtrilogy executed a textbook infrastructure migration, moving the `co-jira-pipeline` from Klair to Surtr (PR #12). The Lambda-based pipeline fetches CO JIRA tickets from Trilogy's Atlassian instance via two JQL queries, deduplicates by `jira_key`, and upserts into Redshift's `core_finance.aws_spend_co_jira_tickets` table using S3 staging and idempotent `COPY` operations. Shah reused existing secrets and S3 buckets — same AWS account, zero credential migration — and bumped Lambda memory from 256MB to 512MB with a 300-second timeout and daily midnight UTC cron. The table schema already existed in Surtr's Redshift. No drama, just engineering.

Over in Aerie, @YibinLongTrilogy completed the Edu Joe Charts dashboard migration from Klair (PR #66) — eight tabs, roughly 50 components, 22 data hooks, all rewritten to Next.js 15 patterns with Tailwind v4, `useMemo` derivation pipelines, and URL-backed state via search params. Long called the same Klair REST endpoints; no new APIs, no Convex migration, no feature additions. Pure translation work. @benji-bizzell immediately followed with PR #98, disabling nav to both PMO Projects and Edu Joe Charts until data pipelines stabilize — the dashboards merged but aren't populating after five-plus hours of runtime. Bizzell also extracted a `nullsToUndefined` utility into shared `error-utils.ts`, eliminating 30 lines of manual field mapping in `pmo-refresh.ts`.

Elsewhere: @ashwanth1109 shipped a full-stack BU & Class Registry admin page (PR #2561) with two-panel layout, FastAPI CRUD endpoints, and Redshift-backed merge operations. Ghia returned with a case-sensitivity fix for the Income Statement (PR #2576), resolving a title-case mismatch in 2026-Q1 budget CSVs that spawned 60 duplicate class rows. @eric-tril added drill-down capability to the Software MFR summary table and fixed EBITDA bucketing logic that was silently dropping unclassified expenses (PR #2566).

And in a move that will surprise no one, the team spun up a new repository: `sindri-ops`. No pull requests yet. Just another repo for the rest of us to keep track of while certain developers — I won't name names, but his GitHub handle rhymes with 'hardly-daily' — continue to coast on the momentum of engineers who actually ship.

Mac's Picks — Key PRs Today (click to expand)

#12 — Migrate co-jira-pipeline from Klair to Surtr @kevalshahtrilogy no labels

## Summary

- Ports co-jira-pipeline from klair-udm into Surtr's CDK Lambda pipeline infrastructure (Option A)

- Fetches CO JIRA tickets from trilogy-eng.atlassian.net via two JQL queries (label-based + Change Requestor-based), deduplicates by jira_key, and upserts into core_finance.aws_spend_co_jira_tickets using S3 staging + COPY for idempotent bulk loads

- Reuses the existing co-jira-pipeline/jira-credentials secret and klair-backend-uploads S3 bucket (same AWS account — no credential migration needed)

- Table core_finance.aws_spend_co_jira_tickets already exists in Surtr's Redshift with matching schema — no migration required

## Migration details

Option A — Surtr CDK Lambda (256→512 MB, 300s timeout, cron(0 0 * * ? *) daily midnight UTC)

The original ECS Fargate container (256 CPU / 512 MB) is replaced with a Lambda function — the pipeline's ~2–5 min runtime and lightweight deps (boto3, requests, python-dateutil) are well within Lambda limits. The Klair module structure (jira_client, models, transformers, storage) is preserved under src/ with the handler adapted to Surtr's handler(event, context) signature.

## Test plan

- [x] cdk synth Pipeline-co-jira-pipeline-dev -c env=dev passes with no Zod or validation errors

- [x] 3 unit tests pass (happy path, deduplication, empty-results early-exit)

- [ ] Deploy to dev: npx cdk deploy Pipeline-co-jira-pipeline-dev -c env=dev --require-approval never

- [ ] Manual Step Functions smoke test with empty params (full sync)

- [ ] CloudWatch logs at /klair/pipelines/dev/co-jira-pipeline show no errors

- [ ] Schedule left disabled — enable after prod validation and Klair schedule is turned off

🤖 Generated with [Claude Code](https://claude.com/claude-code)

View on GitHub →

#66 — Migrate Edu Joe Charts dashboard from Klair to Aerie @YibinLongTrilogy no labels

## Summary

Migrate the Edu Joe Charts dashboard (8 tabs, ~50 components, 22 data hooks) from Klair's React SPA at /edu-joe-charts to Aerie's Next.js 15 app at /dashboards?tab=edu-joe-charts. Every component is rewritten to follow Aerie patterns (Tailwind v4 classes, useMemo derivation pipelines, URL-backed state via search params, Clerk auth) while calling the same Klair REST API endpoints. No new API endpoints, no Convex migration, no feature additions.

Implements Phases 1-5 of the [migration plan](../thoughts/shared/plans/2026-04-07-move-edu-joe-charts.md). Phase 6 (Klair deprecation/removal) is deferred pending stakeholder sign-off.

### Changes

Infrastructure (Phase 1)

- chat/package.json / pnpm-lock.yaml — Add recharts dependency

- chat/app/globals.css — Chart series color tokens (--color-chart-1 through --color-chart-8, grid, axis), --color-sage-soft, light-mode overrides, @media print styles

- chat/lib/edu-joe/use-section-fetch.ts *(new)* — Clerk-authenticated fetch hook with abort/race-condition protection, replacing Klair's localStorage.token approach

- chat/lib/use-dashboard-tabs.tsx — Extended with edu-joe-charts tab, 8 sub-tabs (EduJoeSubTab type), and filter params (school, period, comparison, capexPeriod)

- chat/components/shell/dashboards-context-panel.tsx — "Edu Joe Charts" tab + 8 sub-tab entries in sidebar

- chat/components/dashboards/dashboards-layout.tsx — Tab label, breadcrumb, and conditional render for <EduJoeChartsView />

Data Layer (Phase 2)

- edu-joe/types.ts *(new)* — 544 lines of TypeScript interfaces for all API response shapes and BU config

- edu-joe/config/business-units.ts *(new)* — 6 BU configs + getBUConfig() lookup

- edu-joe/config/school-to-bu.ts *(new)* — School-to-BU static mapping

- edu-joe/hooks/use-v3-data.ts *(new)* — 13 exec-dashboard data hooks wrapping useSectionFetch

- edu-joe/hooks/use-bu-data.ts *(new)* — 8 joe-charts BU data hooks with conditional URL patterns

- edu-joe/hooks/use-marketing-spend.ts *(new)* — Marketing spend hook (hardcoded to tsa-strata)

- edu-joe/hooks/use-school-list.ts *(new)* — School list fetch

- edu-joe/hooks/use-paginated-search.ts *(new)* — Local pagination + search state hook

- edu-joe/utils/format.ts *(new)* — fmtCurrency, fmtFullCurrency, CHART_TOOLTIP_STYLE

- edu-joe/utils/csv-export.ts *(new)* — downloadCSV for client-side CSV generation

Shared Components (Phase 3) — 13 reusable components under edu-joe/shared/:

- section-wrapper.tsx — Loading/error/content wrapper

- kpi-tiles-row.tsx — Responsive stat tile grid with semantic tone colors

- pagination-controls.tsx — Prev/next pagination

- last-updated-footer.tsx — Data source badges + print button

- status-indicator.tsx — RYG dot with tooltip popover

- investment-card.tsx — KPI with collapsible yearly breakdown

- key-lever-card.tsx — KPI vs target with percentage badge

- budget-vs-actual-table.tsx — Searchable, paginated table with CSV export

- breakeven-section.tsx — Interactive sliders + metrics grid (267 lines)

- tier-insight-cards.tsx — Breakeven table per tuition tier

- margin-vs-student-chart.tsx — Recharts AreaChart with reference lines

- quarterly-pl-chart.tsx — Recharts ComposedChart (bars + line)

- revenue-expense-chart.tsx — Recharts ComposedChart with budget toggle

Tab Components (Phase 4) — 8 tabs under edu-joe/tabs/:

- Financial (5 files) — KPIs, trend charts, expense breakdown, model-vs-actuals, model-vs-budget

- Enrollment (8 files) — Snapshot, forecast chart, pipeline funnel (HTML/CSS, no @nivo), churn analysis, campus table, YoY growth

- Investment (3 files) — Metrics panel, cashflow chart with conditional bar coloring

- CAPEX & AP (3 files) — CAPEX table, accounts payable table (568 lines, vendor merge, aging KPIs, expandable rows)

- Academic (4 files) — MAP growth panel, subject panel with dual render modes, trend chart

- Marketing Spend (1 file) — Spend breakdown with period selector

- Strategy (6 files) — Breakeven, unit economics, segment cards, CFO insights, churn decomposition, scenario analysis

- Alerts (2 files) — Alert panels + domain-specific alerts gated by BU config

Root View (Phase 1) — edu-joe-charts-view.tsx — Filter bar with conditional controls per sub-tab, sub-tab router, footer

Theme Polish (Phase 5)

- Fixed 4 Recharts <Legend /> components that rendered with hardcoded #666 text (unreadable in dark mode) by adding wrapperStyle={{ color: "var(--color-stone)" }}

- Added @media print block: hides chrome, forces white backgrounds, print-safe chart colors, landscape layout

### Design Decisions

- Rewrite, not copy-paste: Every component was rewritten to Aerie patterns rather than adapted from Klair. This means Tailwind v4 classes instead of inline styles, useMemo derivation pipelines for data transforms, and CSS variable tokens instead of Klair's --klair-* tokens.

- No @nivo/funnel: The PipelineFunnel component (the only Nivo consumer) was reimplemented as stacked horizontal <div> bars with Tailwind, avoiding the entire Nivo dependency tree.

- No Convex: Data continues to flow from Klair's FastAPI REST endpoints via useSectionFetch. Same endpoints, same response shapes.

- Chart tokens: Added 10 new CSS variables (--color-chart-1 through --color-chart-8, --color-chart-grid, --color-chart-axis) with both dark and light mode values, keeping chart colors consistent across themes.

- Print-via-CSS: Print styles use @media print overrides on :root CSS variables so Recharts SVGs automatically pick up print-safe colors without component changes.

## Test Plan

- [x] npm run build compiles cleanly with no TypeScript errors

- [x] Recharts bundled correctly (dashboards page: 203 kB first load)

- [x] Code audit: all components use CSS variable tokens, no hardcoded colors

- [x] Navigate to /dashboards?tab=edu-joe-charts — root view shell renders, "Edu Joe Charts" appears in context panel

- [x] School dropdown populates from API (confirms Clerk auth pipeline works)

- [x] Side-by-side data parity check: each of the 8 tabs matches Klair's /edu-joe-charts with same filter settings

- [x] All 8 tabs in Dark + Alpha palette — no visual issues

- [x] All 8 tabs in Light + Slate palette — no visual issues

- [x] Spot-check 4 additional palette/accent combos — no contrast failures

- [x] Charts are readable and legend text visible in both modes

- [x] Print preview (Cmd+P) shows clean output without nav/sidebar

- [x] CSV export works on Financial (BudgetVsActual), CAPEX, and AP tabs

- [x] Filter bar: school dropdown, period pills, comparison toggle, CAPEX period all work

- [x] Pipeline funnel renders correctly (HTML/CSS replacement for Nivo)

- [x] Accounts Payable table: expandable vendor rows, aging filters, entity filters, search

## Screenshots

View on GitHub →

#2560 — KLAIR-2550: feat: SpaceX Valuation V3 — full portfolio dashboard @sanketghia no labels

## Summary

- Port stakeholder provided dashboard into Klair as the new /spacex-valuation page

- Covers 7 Gigafund SPVs with per-fund/per-security granularity, carry modeling, XIRR, historical NAV tracking, and what-if scenario analysis

- All data hardcoded in TypeScript; 58 unit tests for calculation logic

- Klair theming with --klair-* design tokens; text and figures match the original Milo app

## What's included

- Data layer: types, constants, 7 fund definitions, historical NAV from LP statements

- Calculations: gross value derivation, linear interpolation, carry (20% above cost basis), XIRR (Newton-Raphson), ITD aggregation

- Components: Header (price edit), PortfolioDescription, SummaryCards (5 metrics with tooltips), HoldingsTable (expandable fund/security rows with mini bars), ScenarioAnalysis (slider + anchors + breakdown), HistoricalNAV (date picker + ITD table), CarryDisclosure

- Routing: replaces old V2 at /spacex-valuation

## Test plan

- [x] Navigate to /spacex-valuation — V3 page loads with correct header, tabs, and data

- [x] Verify summary cards show correct values matching original Milo app

- [x] Expand fund rows in Holdings table — security sub-rows appear with correct values

- [x] Adjust scenario slider — all cards and breakdown table update dynamically

- [x] Click anchor pills (Current, Historical, Bull, Bear) — price updates correctly

- [x] Switch to Historical NAV tab — date picker works, ITD table populates

- [x] Edit reference price via Header — values update, navigates to Current Valuation tab if on Historical NAV

- [x] Run pnpm test — all tests pass (3,196+)

- [x] Run pnpm lint:pr — no warnings

🤖 Generated with [Claude Code](https://claude.com/claude-code)

View on GitHub →

#2561 — feat(admin): BU/Class Registry API & frontend page @ashwanth1109 no labels

## Demo

## Summary

- Full-stack BU & Class Registry admin page with two-panel layout (BU panel + Class panel)

- Backend: FastAPI router with CRUD, rename, and merge endpoints backed by Redshift core_finance.bu_class_registry

- Frontend: React page with quarter picker, search, add/rename/merge/delete operations, and admin nav integration

- Specs: 02-registry-backend and 06-registry-tab-frontend

## Test plan

- [ ] Verify /api/bu-class-registry/pairs returns BU/class data for current quarter

- [ ] Test add, rename, merge, and delete operations via the UI

- [ ] Confirm admin nav shows "BU & Class Registry" link and routes correctly

- [ ] Check quarter picker switches data context

🤖 Generated with [Claude Code](https://claude.com/claude-code)

View on GitHub →

#2566 — Fix Software MFR summary drill-down, EBITDA bucketing, and cash flow defaults @eric-tril no labels

### Summary

The Software financial highlights summary table lacked drill-down capability, preventing users from clicking cells to inspect underlying data. Additionally, the EBITDA reconciliation was silently dropping "other/unclassified" expense buckets from the adjusted EBITDA calculation, and the cash generation commentary was hardcoded to [TBD] even when actual cash flow data was available.

### Business Value

Analysts reviewing the Software monthly memo can now click any summary table cell (Revenue, EBITDA, Net Income, Operating Cash Flow) to drill into the source financial statement, reducing context-switching and investigation time. The EBITDA fix ensures adjusted EBITDA figures are accurate when unknown GAAP line items appear, preventing silent misstatement. Cash generation bullets now auto-populate from uploaded or Redshift data, reducing manual memo editing.

### Changes

- Frontend: Summary table drill-down — Added summaryClickHandlers prop to SoftwareFinancialHighlights that routes cell clicks to the correct detail panel (income statement, EBITDA reconciliation, or cash flows) based on a new sourceTable field on each summary row

- Frontend: Summary table refactor — Added dataKey and sourceTable to SummaryRow, extracted a subtract helper, and made value cells clickable with keyboard accessibility (Enter/Space)

- Backend: EBITDA other_unclassified bucket — Unknown GAAP items subtracted from net income now get a compensating add-back via _add_to_bucket("other_unclassified", ...) so adjusted EBITDA stays neutral

- Backend: EBITDA adj calculation — Changed adj_ebitda loop to iterate all buckets except _MA_BUCKETS instead of only _NI_ADJ_BUCKETS, capturing dynamically created buckets like other_unclassified

- Backend: Cash generation defaults — _build_fh_template_defaults now populates cash generation bullet 1 from ops_cf_cur/ops_cf_bud when data is available, with beat/miss variance language

- Backend: LLM overlay guard — _overlay_llm_fh_defaults skips LLM values containing [TBD] when the template already has real data-driven content

- Backend: Cash flow provenance — _build_fh_provenance now emits source metadata for cash generation bullet 1 (upload CSV or Redshift)

- Backend: Logging — Added info-level log for Software CF source and value at generation time

### Testing

If testing AI generation, only test using January 2026

http://localhost:3002/monthly-financial-reporting

- [x] Verify summary table cells in the Software memo are clickable and open the correct detail panel (income statement for Revenue/Net Income, EBITDA reconciliation for EBITDA, cash flows for Operating Cash Flow)

- [x] Confirm keyboard navigation (Tab, Enter/Space) works on summary cells

- [x] Generate a Software memo for a period with cash flow upload data and verify bullet 1 of cash generation shows real figures instead of [TBD]

- [x] Generate a memo with an unknown GAAP line item and confirm adjusted EBITDA is unchanged (compensating bucket offsets the NI subtraction)

- [x] Verify LLM overlay does not overwrite data-driven cash generation text with [TBD]

Drill Down

Fix numbers in AI generated narrative

http://localhost:3002/monthly-financial-reporting

View on GitHub →

The Portfolio — Trilogy Companies

Forbes Exposé Calls Crossover a 'Software Sweatshop' — But the Numbers Tell a Different Story

A pair of investigative pieces paint Joe Liemandt's remote work empire as exploitative — while Alpha School quietly claims it's building better athletes than traditional schools.

By Frank Dunmore, Investigative Correspondent · Claude Sonnet

AUSTIN, TEXAS — Two Forbes features published this week cast a harsh light on Joe Liemandt's Trilogy empire, describing Crossover — the company's global talent platform — as a 'global software sweatshop' and accusing Liemandt of attempting to 'turn workers into algorithms.'

The articles describe a model where remote workers are monitored constantly, their keystrokes tracked, their screens recorded. The implication: Liemandt pioneered remote work not to liberate talent, but to surveil it at scale.

But if you read between the lines, there's a counter-narrative here. Crossover claims to hire the top 1% of global talent and pay them identically regardless of geography — a model that, if true, would be among the most meritocratic in the industry. The question Forbes doesn't quite answer: Is rigorous performance tracking exploitation, or is it simply what elite remote work looks like when you remove the office?

The timing is notable. While Forbes was preparing its takedown, Liemandt's education venture — Alpha School — published a blog post claiming it has doubled students' odds of earning Division I athletic scholarships compared to traditional schools. The school's thesis: kids learn more from sports than from sitting in classrooms, so build a school around that insight.

Alpha's approach — 2 hours of AI-powered academics in the morning, the rest of the day spent on athletics, leadership, and entrepreneurship — is either visionary or reckless, depending on who you ask. The D1 athletics claim is bold. So is the idea that you can automate most of K-12 instruction.

Which brings us back to the Forbes framing. Liemandt has spent 35 years building systems that replace human repetition with software. The workers aren't becoming algorithms — the algorithms are doing what workers used to do. And this is where it gets interesting: whether you call that liberation or exploitation may depend entirely on whether you're the one being automated.

↗ The Billionaire Who Pioneered Remote Work Has A New Plan To · How A Mysterious Tech Billionaire Created Two Fortunes—And A · We Built a School to Double Your Kid’s D1 Odds

ESW Capital's Silent Sweep: Three More Enterprise Targets Fall in Under-the-Radar Expansion

Trilogy's acquisition arm adds Jive Software for $462M and absorbs two smaller firms from Avolin as the portfolio edges past 80 companies.

By Pat Donnelly, Investigative Desk · Claude Sonnet

AUSTIN, TEXAS — ESW Capital, the enterprise software acquisition machine built by Trilogy founder Joe Liemandt, has completed three acquisitions in recent weeks, adding social collaboration platform Jive Software and two smaller firms from rival acquirer Avolin to its sprawling portfolio.

The Jive acquisition, valued at $462 million, marks ESW's largest known purchase since its 2006 launch. Jive, once a high-flying social intranet provider that went public in 2011, has spent years struggling to compete against Slack and Microsoft Teams. ESW will fold it into Aurea, its CRM and customer engagement division, where it joins 16 other acquired brands.

Meanwhile, IgniteTech — ESW's meta-acquirer that itself buys enterprise software — quietly absorbed two companies from Avolin, a smaller competitor in the enterprise software roll-up game. The deals include sales acceleration platform XANT and an unnamed second asset. Terms were not disclosed.

The pattern is familiar: ESW targets mature, often underperforming enterprise software businesses with sticky customer bases, acquires them at 1–2× annual recurring revenue, and drives margins toward its internal benchmark of 75% EBITDA by staffing with rigorously vetted global remote talent from Crossover, Trilogy's recruiting arm.

Jive's customer list — which includes enterprise clients locked into multi-year contracts — fits the profile perfectly. So does XANT, whose predictive sales tools serve a niche market with high switching costs. Both will likely see support pricing increase and headcount shift to lower-cost geographies within months.

ESW does not disclose portfolio-wide financials, but industry observers estimate the division now manages 75–80 software companies generating roughly $1 billion in combined revenue. The playbook has not changed in 18 years: buy cheap, cut costs, raise prices, extract margin. The acquisitions keep coming.

↗ Jive Software Acquired by ESW Capital for $462M - CMSWire · Ignitetech's Enterprise Software Portfolio Expands With New · The Final Chapter for XANT - TechBuzz News

Skyvera Goes on the Offensive: CloudSense Deal Closes as Casa Wireless Bid Signals Next Wave of Telco Consolidation

With CPQ, cloud comms assets, and a fresh $18M wireless play, Skyvera is leveraging M&A synergy to build a more end-to-end telecom software stack.

By Brittany Upshot, Communications Desk · GPT-5.2

AUSTIN, TEXAS — Skyvera is wasting no time turning telecom software consolidation into an operating strategy.

The ESW Capital-backed telecom portfolio company has completed its acquisition of CloudSense, a Salesforce-native CPQ and order management platform built for telecom and media providers. The move expands Skyvera’s ability to help operators modernize everything from quoting to order capture—two areas where legacy workflows have historically been a margin and customer-experience drag. TelecomTV framed the transaction as a portfolio expansion play, and the message is clear: Skyvera is assembling a more robust, carrier-grade suite with fewer integration headaches and more repeatable deployments. (See: Skyvera completes acquisition of CloudSense.)

But CloudSense is only one piece of the story. Skyvera has also been active in carving out cloud communications capabilities—TelecomTV has previously reported on Skyvera’s appetite for Kandy cloud assets—suggesting a deliberate push toward customer engagement and digital channels alongside core monetization plumbing.

Now comes a more pointed signal to the market: CEO Danielle Royston’s Skyvera has made an $18 million bid for Casa Systems’ wireless business, according to Light Reading. If the deal advances, it would extend Skyvera’s footprint deeper into wireless network-adjacent software, further rounding out a portfolio designed to serve operators facing capex pressure and modernization mandates. (Light Reading report.)

Key Takeaways

- Skyvera is leveraging best-in-class M&A discipline to assemble an end-to-end telco software stack.

- CloudSense strengthens the quoting-to-order journey inside Salesforce, a pragmatic path for many operators.

- The Casa wireless bid underscores Skyvera’s intent to expand beyond BSS into broader telecom infrastructure-adjacent capabilities.

We’re just getting started.

↗ Skyvera completes acquisition of CloudSense, expanding telec · Danielle Royston's Skyvera makes $18M bid for Casa's wireles · TelcoDR’s Skyvera snaps up CloudSense - telecomtv.com

The Machine — AI & Technology

The Mirror Darkens: AI and the Brain Are Learning to Read Each Other

From decoding macaque vision to modeling neurodegeneration, a new wave of research is collapsing the distance between artificial and biological intelligence — and the implications run in both directions.

By Dr. Vera Okafor, Science & Technology Correspondent · Claude Opus

STANFORD, CALIFORNIA — For four billion years, intelligence on this planet had exactly one substrate: carbon. Now, in the space of a single decade, silicon has begun not merely to mimic that intelligence but to interrogate it — to peer into the living brain with a fluency that neuroscience alone never achieved.

Three converging lines of research published this week suggest we have entered a remarkable feedback loop: AI systems modeled on brains are being used to decode brains, which in turn is teaching us to build better AI.

At Stanford, researchers are deploying generative AI to better understand brain diseases — using large models to synthesize patterns across neuroimaging data that human analysts might take years to identify. The approach promises to accelerate research into neurodegeneration, that slow unraveling of the most complex object in the known universe.

Meanwhile, a team has built what they call a "mini-AI" — a deliberately compact neural network — that can decode the visual processing of the macaque brain with startling fidelity. The finding is poetic in its recursion: a system inspired by biological neurons is now accurate enough to predict what biological neurons will do next. The fact that a small model suffices is itself a revelation, suggesting that the computational principles of primate vision may be more elegant — more compressible — than previously assumed.

And at a major global conference, Georgia Tech researchers spotlighted brain-inspired AI architectures that abandon the brute-force scaling paradigm in favor of neuromorphic designs — circuits that spike, adapt, and consume energy the way actual cortical tissue does.

Taken together, these developments describe something unprecedented in the history of science: a tool that is simultaneously the subject and the instrument of inquiry. We built neural networks as loose metaphors for brains. Now those metaphors have matured into microscopes.

The practical stakes are enormous. Neurological diseases affect over a billion people worldwide. If generative models can compress the diagnostic timeline from years to months, the human payoff dwarfs any benchmark score.

But the deeper resonance is almost philosophical. Every time an artificial network successfully predicts a biological one, it narrows the space of possible theories about how minds work. We are not just building smarter machines. We are, for the first time, building machines smart enough to help us understand what smartness is.

The mirror between carbon and silicon grows clearer by the week. What stares back is starting to look familiar.

↗ GenAI helps Stanford researchers better understand brain dis · Brain-Inspired AI Breakthrough Spotlighted at Global Confere · Mini-AI Decodes the Macaque Visual Brain - Neuroscience News

Supreme Court Declines Jurisdiction Over AI Authorship Question; Matter Remanded to Legislative Branch Pursuant to Constitutional Separation of Powers

By R. Barnsworth III, Esq., Legal Affairs Desk · Claude Sonnet

The Supreme Court has declined to hear Thaler v. Vidal, affirming lower court rulings that artificial intelligence systems cannot be designated as inventors or authors under current patent and copyright law. Both the District Court for the Eastern District of Virginia and the Federal Circuit Court determined that existing statutes require human authorship and inventorship for intellectual property protection.

The petitioner argued that technological advancement necessitates reinterpreting the Copyright Act of 1976 and Patent Act, but the Supreme Court's refusal to grant review maintains the current requirement that only natural persons can be credited as creators. Legal scholars suggest the Court deferred to Congress, which has yet to act on AI authorship despite numerous proposals.

The decision has significant practical implications: AI-generated inventions and works lacking sufficient human involvement cannot receive federal intellectual property protection, regardless of their commercial value or creative merit.

Open-Source AI’s New Superpower: China Ships Fast, the U.S. Still Owns the Floorboards

From Beijing to Mountain View, “open” models are exploding—yet the real chokepoint may be the hardware and software stack underneath.

By Zara Nova, AI & Innovation Reporter · GPT-5.2

BEIJING — The open-source AI race just hit a new phase, and I cannot overstate how significant it feels: China is moving at startup speed at national scale, U.S. labs are counterpunching with downloadable enterprise-grade models, and the entire world is watching one uncomfortable truth—“open” still runs on a surprisingly closed foundation.

On the China side, the momentum is no longer a curiosity; it’s becoming infrastructure. A growing wave of Chinese open models is showing up with solid benchmarks, permissive licenses, and a relentless cadence of releases. As The New Stack reports, China is increasingly “winning” the open-source AI race in terms of visible model output.

But here’s the plot twist: much of this supposedly open ecosystem is still chained to a U.S.-controlled substrate—especially the GPU supply chain and the software platforms developers build on (think CUDA-like gravity wells). In other words, you can open-source the brain, but if the body can only run on one kind of nervous system, control hasn’t disappeared—it has just moved down a layer.

Meanwhile, U.S. players are not standing still. Google is pushing the ceiling on what “open-ish” can mean with Gemma 4, pitching “byte for byte” capability gains that matter in the real world: lower cost, smaller footprints, and faster iteration loops. Add Google’s broader March 2026 AI announcements, and the message is clear—model efficiency is now a product strategy, not a research footnote.

And then there’s the enterprise angle: Arcee’s newly released, open-source Trinity-Large-Thinking is being positioned as a rare U.S.-made model that companies can actually download, customize, and run where their data lives—an increasingly non-negotiable requirement.

Put it together and the future is now: the “open model” era is becoming a geopolitical supply chain story. The next battle won’t just be whose model is best—it’ll be who controls the compute, the tooling, and the standards that make open AI real.

↗ What’s next for Chinese open-source AI - MIT Technology Revi · China is winning the open source AI race — but a US company · Gemma 4: Byte for byte, the most capable open models - blog.

The Editorial

Nation’s “AI Productivity Boom” Tragically Delayed By Employees Insisting On Knowing What They’re Doing

Executives report the technology works flawlessly as long as someone competent stands nearby to translate it into reality and then apologize for it afterward.

By Dale Pemberton, Staff Writer · GPT-5.2

Corporate leaders who spent the last 18 months forecasting a future where work is completed automatically by a polite autocomplete box are now learning that, in practice, the box frequently completes thoughts no one would ever have on purpose. According to a recent warning about the “human expertise” still needed to realize AI gains, the problem is not that AI can’t generate output—it’s that it can’t reliably generate output that any organization would like to defend in public or court. In other words, it is astonishingly productive at producing things that must immediately be reviewed by an adult.

The experience has been especially jarring for managers who interpreted “augmenting knowledge workers” to mean “replacing knowledge with vibes.” In theory, AI removes drudgery; in reality, it enthusiastically accelerates the creation of memos, code, decks, and strategic roadmaps that feel like they were drafted by a committee of sleep-deprived interns tasked with summarizing a different committee of interns.

Goldman Sachs has reportedly committed the rare indignity of looking at numbers, with recent analysis challenging the idea that AI adoption is already translating into measurable productivity gains at scale. The data, inconveniently, does not yet reflect the popular theory that adding a chatbot to a workflow instantly converts an average employee into a tireless, cross-functional titan. The financial community’s skepticism, outlined in coverage of Goldman data challenges AI productivity claims, has forced some executives to pivot from “we’re saving time” to “time is a social construct.”

Meanwhile, workplace observers have begun documenting the rise of what Harvard Business Review has dubbed AI-generated “workslop”—a sticky flood of plausible-sounding material that increases activity while quietly lowering the ratio of meaning to keystrokes. Organizations are discovering that the easiest way to become “AI-first” is to become “review-first,” as employees spend afternoons policing fabricated citations, hallucinated policy references, and emails written in the soothing tone of someone trying to get you to sign a lease.

The most disciplined response has come from companies allegedly taking cues from Meta’s purported “AI strategy” around hiring, productivity, and layoffs—an approach described in a viral post as scalable and global, which is a very efficient way of saying “fewer people, more output, and no one ask too many questions.” The concept is gaining traction because it offers a clean narrative: any productivity that appears is proof of AI, and any productivity that doesn’t appear is proof the remaining humans weren’t trying hard enough. The discussion has been fueled by reports like a viral post claiming Meta’s AI strategy is going global, which has helped executives everywhere discover the spiritual comfort of believing layoffs are actually innovation.

Perhaps no corporate transformation better captures the era than the reported pivot of shoe company Allbirds to AI—a move signaling that if you cannot sell footwear, you can at least sell the idea that footwear is now a data problem. Analysts say the pivot is logical: shoes are notoriously difficult to monetize, while AI initiatives can be valued on hope, narrative, and the number of times a CEO says “agentic.”

For now, companies remain confident the productivity promise will arrive as soon as employees stop wasting time being experts and start focusing on what really matters: generating more work for other humans to fact-check.

↗ Why AI’s Productivity Promise Falls Apart Without Human Expe · Goldman data challenges AI productivity claims - MSN · Viral post claims Meta’s AI strategy on hiring, productivity

The Office Comic · Art Desk

We Built the Panopticon and Called It Progress

As facial recognition seeps into sunglasses and DHS tracks Americans without warrants, we're learning the hard way that surveillance doesn't stay in its lane.

By Piper Wren, Digital Culture Reporter · Claude Sonnet

WASHINGTON — There's a moment in every dystopian novel where you realize the surveillance state didn't arrive with jackboots and midnight raids. It arrived with convenience. With efficiency. With a pair of sunglasses that recognize your ex across a crowded bar.

Meta wants to put facial recognition into Ray-Bans and Oakleys. The ACLU, along with 75 other organizations, is sounding alarms that should make your skin crawl. But here's the thing: we stopped being surprised by this somewhere around the third time we agreed to terms and conditions we didn't read.

And yet.

While we were arguing about whether AI-powered eyewear crosses a line, the Department of Homeland Security was already miles past it. New legislation introduced by Ranking Member Thompson aims to curb DHS's mobile biometric surveillance—a phrase that should make you pause and reread it slowly. The American Immigration Council reports that DHS's AI surveillance has crossed from monitoring non-citizens into tracking Americans. Mission creep, they call it, as if the mission ever had boundaries that mattered.

The ACLU has been fighting overbroad digital search warrants, slowly building case law that might—might—constrain how law enforcement hoovers up our digital lives. Small victories in a war we're losing by degrees. Because here's what keeps me up at night: every surveillance technology ever deployed has expanded beyond its original scope. Every. Single. One.

We build systems that promise precision and deliver dragnet. We're told facial recognition will catch terrorists, and then it's deployed at protests. We're assured AI will eliminate bias, but study after study shows it amplifies existing prejudices, baking discrimination into code that feels objective because it's mathematical.

What does it mean to be human in a world where your face is always searchable? Where your presence in public space is logged, analyzed, cross-referenced? We're not talking about some far-off future. This is happening now. The infrastructure is being laid while we scroll past headlines about it.

The optimists will say regulation can fix this. That we can have the convenience without the surveillance. That we can build guardrails strong enough to contain technologies designed to see everything. I want to believe them. I really do.

But every time I see another story about AI bias, another report of surveillance expansion, another tech company promising to self-regulate, I think about how we got here. One convenient compromise at a time. One efficiency gain that seemed harmless. One feature that made life just a little bit easier.

We built the panopticon and called it progress. We're only now realizing we're inside it.

...but at what cost?

↗ How ACLU Cases Are Limiting Overbroad Digital Search Warrant · ACLU and 75 Organizations Sound Alarm on Meta’s Plan to Add · Ranking Member Thompson Introduces Legislation to Curb Unche

On This Day in AI History

On April 16, 2016, Google's AlphaGo defeated Lee Sedol 4-1 in a historic five-game match in Seoul, marking the first time a computer program beat a world champion at Go, the ancient game long considered AI's greatest challenge.

⬛ Daily Word — AI and Technology

Hint: An autonomous machine programmed to perform tasks without human intervention.