Research that stays true

Report #1

How Large Language Models Actually "Think": Tokens, Embeddings & Transformers

Report #3

Will AI Replace Software Engineers by 2030? What the Research Says

AI will not replace software engineers by 2030, but it will replace engineers who refuse to use AI. The research points to a shift from writing code to specifying, reviewing, and orchestrating it — raising output per engineer while changing what the job rewards.

Report #4

The Hidden Jobs AI Will Create (Not Destroy)

Every major automation wave destroyed some jobs and created more. AI is already spawning new roles — AI engineers, evaluators, safety and alignment specialists, agent orchestrators, and AI-augmented versions of old jobs — often paying above the ones they replace.

Report #5

What Happens When Every Worker Has an AI Agent?

When every knowledge worker has a capable AI agent, the unit of work shifts from "tasks done" to "outcomes directed." Output per person jumps, org charts flatten, and the scarce skill becomes deciding what to do — not doing it.

Report #6

The Rise of Autonomous AI Agents: The Next Big Shift

Autonomous AI agents — systems that plan, use tools, and complete multi-step goals with little supervision — are the next platform shift after chatbots. This report explains how they work, where they already deliver value, and where they still fail.

Report #7

AI Agents Are Quietly Rebuilding the Internet

The web was built for humans clicking pages. AI agents browse, transact, and call APIs on our behalf — reshaping search, advertising, SEO, and how sites are designed. This report explains the shift from a human-first to an agent-first internet.

Report #8

Can AI Agents Run an Entire Business Alone?

AI agents can already run large parts of a business — marketing, support, ops, and code — but not the whole thing unsupervised. This report maps which functions agents handle today, where they break, and what the "AI-run company" really looks like in practice.

Report #9

The End of Manual Workflows in Tech Companies

The repetitive glue work that fills knowledge workers' days — copy-pasting between tools, status updates, manual data entry — is being automated away by AI and agents. This report shows which workflows go first and how teams should restructure.

7 min read PDF

Report #10

AGI Timeline: What the Research Actually Predicts

Predictions for artificial general intelligence range from "a few years" to "never." This report synthesizes expert surveys, scaling-law research, and historical forecasting accuracy to give an honest, evidence-based picture — and why the exact date matters less than people think.

Report #11

The Real Cost of Running AI at Scale (What Nobody Tells You)

AI looks cheap per query and expensive per company. This report breaks down the true economics — inference vs training costs, GPU scarcity, energy, and why most "AI features" quietly lose money — and how to build AI products that are actually profitable.