BRENNEN

Industries / Finance and business services / Legal and professional services

Consulting

Consulting firms are hired to study an organization's problem, in strategy, operations or technology, and recommend what to do, through interviews, analysis, models and reports, and sometimes by helping to carry the advice out.

1886

In Boston, the MIT-trained chemist Arthur Dehon Little and his partner Roger B. Griffin opened a laboratory and office on Milk Street that did chemical analysis and industrial research for clients, much of it on papermaking. The firm calls it the first consulting firm dedicated to improving industrial processes and products, and MIT says the company "is known as the world's first management consultancy.".

adlittle.combetterworld.mit.edu

36

Milestones achieved across 6 domains.

8

To watch for: announced, not yet confirmed outside.

12

Open deliverables. Nobody has shipped them.

In one minute

Where it started

Consulting began as research for hire: a Boston laboratory in 1886, and in 1914 a Chicago firm called The Business Research Service, now Booz Allen Hamilton. In 1911 Frederick W. Taylor described studying the work at Bethlehem Steel and resetting how much each man should do, the model for an outsider who studies a job and changes it. The trade's everyday tools followed: large interview programs from 1928, the growth-share matrix in 1970, the electronic spreadsheet in 1979 and PowerPoint in 1987.

Where it stands

The work is still interviews, spreadsheets, slides and reports, and outside tests now measure how much of it AI can do: in January 2026 the best AI agent finished 22.7% of Mercor's consulting tasks on the first try, and in June 2026 the best model fixed 12% of the faulty workbooks in SpreadsheetBench 2. Randomized trials in India and Mexico found that consulting raised firms' productivity, though about half of the practices adopted in India had been dropped when researchers returned in 2017. Firms have paid penalties and faced inquiries for letting confidential client and government information travel inside them.

What is open

Most open jobs are checks the trade does not yet run on itself: a public scorecard of past market forecasts, tests of whether AI or people catch planted errors in models and find planted facts in interviews, reader tests of slide decks, and a record of what happened to the recommendations governments paid for. Each can be started this month with public records, open-source tools and current AI models, on published or made-up material, without advising a real client.

Domains

AchievedWatch forOpen

Advances to build with

1 research note

Open in consulting

3 of 12 open deliverables

Open · Research and analysis

A forecast scorecard

A public table of market forecasts that consulting and research firms published for a target year that has now passed, each set against the figure that came true. Every row links the forecast, states exactly what was counted, gives the actual figure from an official or audited source, and shows the error.

Built with what exists

the firms' published reports and press releases; official statistics such as US Census Bureau and Eurostat tables; a current language model to pull each forecast's number, unit, scope and target year into a table; a spreadsheet that a second person checks.

Credit · unclaimed

Open · Interviews and fact-finding

Planted facts, two interviewers

A head-to-head test of AI and human interviewers. Role-players act as staff of a made-up business with a problem, such as orders that ship late, and each holds a written list of facts that they reveal only when a question reaches them. AI and human interviewers get the same brief, and graders count how many planted facts each interview uncovers.

Built with what exists

the open-source interview platform of Friedrich Geiecke and Xavier Jaravel (github.com/friedrichgeiecke/interviews) or another current AI model as the AI interviewer; volunteer or paid role-players; a video-call tool; speech-to-text such as ElevenLabs Scribe v3 for voice interviews (see the Note "Fewer word errors"); a scoring sheet and two graders.

Credit · unclaimed

Open · Modeling

Planted errors, people versus AI

A public test of model checking. Take 30 models of the kinds consultants build, such as cost models, business cases and cash forecasts, plant known errors of each common type in copies, and give the same copies to AI reviewers and to experienced human model reviewers. Count what each catches, by error type, and how often each flags a cell that is correct.

Built with what exists

the open-source openpyxl library to plant errors and list every formula; LibreOffice in headless mode to recalculate; AI reviewers such as Claude for Excel and Microsoft's Agent Mode in Excel, plus a scripted agent through an API; Panko's error types (mechanical, logic and omission) as the starting list; each AI reviewer run five times on every file, counting errors caught on every run (see the Note "All twenty times").

Credit · unclaimed

Where it sits