Futurist > Companies creating the future > 13 companies harnessing the power of collective intelligence

 
Graphical representation of Power of Collective Intelligence
 

13 companies harnessing the power of collective intelligence

Last updated November 2026

Collective intelligence is the capacity of a group to reach better judgments, forecasts and decisions than its members could reach alone. The organizations on this list build the mechanisms that make this work: markets that price shared expectations, forecasting platforms that aggregate and score predictions, deliberation tools that let thousands of people reason together, and methods that turn many individual observations into usable insight.

The field has changed substantially since 2021. Prediction markets have moved into mainstream finance. Kalshi raised $1 billion at a $40 billion valuation in October 2026 and reported more than 30 million accounts. Polymarket’s latest round, reported in August 2026, values it at $21 billion.

AI has also become a participant in collective intelligence. Unanimous AI uses AI agents to link small human discussion groups into one large conversation. Metaculus runs a standing benchmark that compares AI forecasting bots with its best human forecasters. In its Spring 2026 tournament the humans still led, by a margin that was not statistically significant. Where AI plays a role, the designs on this list have people and AI working together, with AI handling scale and synthesis while people supply judgment, context and values.

Public deliberation has matured as well. In September 2025 Google’s Jigsaw used AI to help more than 2,400 Americans from all 435 congressional districts find common ground on freedom and equality. The Collective Intelligence Project ran six Global Dialogues on AI in 2025, with more than 10,000 participants in over 70 countries.

The thirteen organizations below are ordered by their current significance to the field.


Kalshi

A US prediction market exchange where people trade contracts on the outcomes of events in sports, economics, politics, commodities and crypto.

kalshi.com

Regulator: Commodity Futures Trading Commission (CFTC)
Valuation: $40 billion post-money after a $1 billion Series G (October 2026)
Latest: November 2026 launch of single-stock perpetual futures on more than 50 companies listed on the New York Stock Exchange and Nasdaq

Kalshi turns a question about the future into a tradable contract whose price reflects what participants collectively expect to happen. In October 2026 the company reported more than 30 million accounts, over 75% of prediction market trading volume and more than $1 billion in annualized revenue from non-sports contracts. Kalshi says sports contracts brought millions of people to the platform and established its liquidity.

Kalshi is extending its exchange beyond event contracts into politics, crypto, economics, commodities and perpetual futures. Its perpetual futures business reached $83 billion in trading volume within a few months of launch. In October 2026 it announced a collaboration with S&P Dow Jones Indices to build prediction market indices.


Polymarket

A prediction market where traders buy and sell shares in the outcomes of events, operating a blockchain-based international platform and a separate regulated US exchange.

polymarket.com

US exchange: Polymarket US, operated by QCX LLC, a CFTC-regulated designated contract market
Valuation: A reported $21 billion in a $1 billion round led by 1789 Capital (reported August 2026, not confirmed as closed)
Latest: August 2026 reports of a new round at a $21 billion valuation, up from $15 billion in April 2026

Polymarket is Kalshi’s main competitor. It runs a blockchain-based international platform alongside Polymarket US, a separate regulated exchange for US traders with markets on sports, politics, crypto, commodities, weather and culture. Intercontinental Exchange (ICE), owner of the New York Stock Exchange, agreed in October 2025 to invest up to $2 billion at a valuation of about $8 billion before the investment.

D.E. Shaw and G Squared invested in Polymarket’s April 2026 round. In August 2026 Bloomberg reported that 1789 Capital would lead a further round with an expected investment of about $300 million. Polymarket is courting institutional traders and is seeking US regulatory approval to offer margin trading.


Metaculus

An online forecasting platform that aggregates community predictions on science, technology, geopolitics and AI, and benchmarks AI forecasting bots against its best human forecasters.

metaculus.com

Model: Community forecasting platform with questions scored as they resolve
AI benchmark: FutureEval, with $175,000 a year in prizes for forecasting bots
Latest: Spring 2026 FutureEval results (January to April 2026) showed Pro Forecasters ahead of the top AI bots, by a margin that was not statistically significant

Metaculus poses questions about future events, collects probability forecasts and scores them as the questions resolve. Its community forecast combines many individual predictions into a single estimate. A team of hand-picked Pro Forecasters provides a high-skill human baseline.

Since 2024 Metaculus has run tournaments in which AI bots forecast the same questions as people. The Spring 2026 tournament involved 173 bots on 297 scored questions. A team of ten Pros beat a team of the top ten bots by an average of 1.25 points per question, a much smaller gap than in earlier seasons. Hobbyists built seven of the top ten bots. These results give organizations evidence on where AI forecasting already adds value and where human judgment still leads.


Good Judgment Inc

The commercial home of the Superforecasters, a network of people selected for consistently accurate forecasting, who provide custom forecasts, a monitoring service and forecasting training to organizations.

goodjudgment.com

Offer: Custom Superforecasts, the FutureFirst monitoring tool and training
Benchmark: Superforecasters lead ForecastBench on real-world event questions, 80.3 to 75.8 against the best AI system (April 2026)
Latest: July 2026 analysis disputing claims that AI has reached parity with Superforecasters, alongside work on its own AI-powered Superforecasting tools

Good Judgment aggregates the judgments of its Superforecasters into probability estimates for clients. Its Superforecasters contributed forecasts to The Economist’s The World Ahead 2026 on topics including the US midterm elections, US tariffs and the Russia-Ukraine conflict. It also publishes forecasting reports on subjects such as AI governance and global health funding.

The company is an active participant in the debate over whether AI now forecasts as well as skilled people. It argues that published comparisons set current AI models against a frozen 2024 sample of Superforecasters working largely alone, on different questions. On ForecastBench questions about real-world events, the best AI system’s error rate was about 50% higher than the Superforecasters’ in April 2026. Good Judgment describes its direction as a hybrid, with AI amplifying what Superforecasters do and people kept in the loop on all client work.


Unanimous AI

Builds AI agents that connect many small human discussion groups into one real-time conversation, so large teams can deliberate and converge on decisions together.

unanimous.ai

Platforms: Swarm and Thinkscape, which supports real-time deliberation in teams of up to 250 people
Customers: Fortune 500 companies, sports teams, hedge funds, the US Department of Energy and the United Nations (company-reported)
Latest: June 2026 study with Carnegie Mellon in which Thinkscape teams of 25 to 30 sports fans picked 62% of 50 NBA games against the spread, compared with 55% for Polymarket

Unanimous AI began with Swarm AI, which models group decisions on the way biological swarms such as fish schools converge on a choice. Its newer Hyperchat AI technology addresses a basic limit of conversation: discussion works best in groups of about four to seven people. Hyperchat splits a large group into small subgroups and gives each an AI agent that carries insights between them. The whole group deliberates as one, while the judgments stay with the people.

In a 2024 study with Carnegie Mellon, individuals averaging an IQ score of 100 performed at a level of 128 when connected through Hyperchat AI. In May 2026 the company won a one-year contract with AFWERX, the US Air Force innovation unit, to integrate Hyperchat AI into Microsoft Teams. Most of the published research is co-authored by the company, so independent replication remains an open question.


Jigsaw (Google)

A Google unit that builds open-source Sensemaking AI tools to help large groups of people deliberate and find common ground.

jigsaw.google

Parent: Google
Flagship deliberation: We the People, with more than 2,400 Americans from all 435 congressional districts (September 2025)
Latest: June 2026 update of its open-source Sensemaking code library on GitHub, which now runs the full pipeline for any organization

Jigsaw’s Sensemaking suite has three parts. An AI interviewer asks each participant tailored follow-up questions. AI summarization groups responses into opinion clusters grounded in the original quotes. A predictive agreement engine drafts statements likely to win broad support and estimates how each participant would vote on them.

In We the People, run with the Napolitan Institute, participants wrote more than 1.6 million words on what freedom and equality mean to them. The process produced 26 summary statements, most of which won more than 80% agreement among participants. Of the participants, 94% said they felt their opinions were represented. Jigsaw designs the tools so that project owners and facilitators review all outputs before they are shared.


The Collective Intelligence Project

Gathers structured public input from around the world and turns it into evaluations that shape how AI systems are built and governed.

cip.org

Focus: Democratic input into frontier AI development
Reach: Six Global Dialogues in 2025, with more than 10,000 participants in over 70 countries
Latest: December 2025 plan to make Global Dialogues standing global infrastructure during 2026

The Collective Intelligence Project (CIP) launched Global Dialogues at the Paris AI Action Summit in February 2025. Each dialogue combines stratified sampling, AI-assisted facilitation and translation, and bridging-based aggregation that captures minority concerns as well as majority views. The platform is built with Remesh, which runs the deliberation, and Prolific, which recruits participants and collects demographic data.

CIP reports that AI labs and governments from India to the UK and Taiwan have used its findings. It also turns public priorities into evaluations that labs and governments can use to steer AI models. The model has people and AI working together at global scale: AI handles translation and synthesis so that thousands of human judgments can inform how AI itself is developed.


Remesh

An AI-assisted platform for live and asynchronous conversations with hundreds or thousands of people, who answer in their own words and vote on one another’s responses.

remesh.ai

Headquarters: Cleveland, Ohio
Uses: Market research, employee listening and public dialogue, including the Collective Intelligence Project’s Global Dialogues
Latest: October 2026 launch of Remy Interviews, one-on-one video interviews moderated by its AI research agent, which researchers can take over at any time

In a Remesh conversation a moderator asks questions and participants respond in free text. Participants then vote on samples of each other’s answers. Remesh’s algorithm and language models predict how each participant would vote on responses they did not see, which reveals where a large group agrees and disagrees.

Remesh runs the deliberation layer of CIP’s Global Dialogues. In September 2026 it published an account of three dialogues that used the platform to gather Cypriot views on the peace process. Since July 2026 teams can work with Remesh data and its AI agent from inside Claude, ChatGPT and other assistants. Its core business is enterprise market research and employee listening.


The Computational Democracy Project (Polis)

The nonprofit behind Polis, an open-source system that uses statistics and machine learning to show where large groups agree and disagree.

compdemocracy.org

Founded: 2018, as a nonprofit continuation of Polis
Status: Verified digital public good, available in more than 25 languages
Latest: Digital Public Goods Alliance evaluation completed in January 2026

In a Polis conversation, participants submit short statements and vote on statements from others. Polis groups participants by their voting patterns and highlights statements that win support across groups, which shows decision-makers where common ground exists. In Taiwan, a Polis process involving taxi drivers, Uber drivers and passengers informed regulation that let Uber operate with licensed drivers.

Polis can now generate narrative reports with large language models, as an optional feature that the organizers of a conversation choose to switch on. It is often combined with other tools: Bowling Green, Kentucky used Polis alongside Jigsaw’s Sensemaker in a citywide consultation. Its code, open data and analysis notebooks are public, so governments, researchers and journalists can run and audit their own deliberations.


Loomio

Open-source software for group discussion and decision-making, built by a worker-owned cooperative in New Zealand.

loomio.com

Founded: 2011
Ownership: Worker-owned cooperative based in Aotearoa New Zealand
Latest: February 2026 case study on how NYC-DSA used Loomio to coordinate 13,000 members during Zohran Mamdani’s 2025 mayoral primary campaign

Loomio gives a group a dedicated space to share context on a question, propose a course of action and vote with reasons. The discussion, the reasoning and the outcome remain searchable afterward. This addresses the loss of context common to decisions made in chat, email or meetings. Groups can also self-host the software and keep control of their data.

Users include the World Resources Institute and the OpenSSL open-source community. The New Zealand teachers’ union NZEI Te Riu Roa used Loomio when 28,000 members debated strike action. Loomio’s contribution to collective intelligence is process design that lets every member take part.


bluenove

A French firm that designs and runs large-scale consultations, using its Assembl platform and AI under human supervision to analyze contributions from thousands of participants.

bluenove.com

Track record: More than 500 projects delivered and 10 million contributions analyzed
Owner: Groupe Alpha
Latest: September 2026 publication of its synthesis of the French national consultation on children’s books and reading

bluenove runs consultations, co-creation processes and facilitated workshops for public and private organizations. Its Assembl platform gathers open-ended input, then uses semantic and generative AI to map themes, surface weak signals and build summaries, with people supervising the analysis. In 2026 it added Assembl Workshops, which transcribes discussions and captures the outputs of in-person sessions.

Its largest projects run at national scale. A French national consultation on health and social protection drew 76,873 citizens and health professionals, who gave 2.29 million answers including 405,007 written responses. bluenove also trains client teams in the use of AI.


The Cynefin Company

A research and strategy firm founded by Dave Snowden that helps leaders make sense of complex situations, using the Cynefin framework and the SenseMaker method for collecting and interpreting people’s own accounts of their experience.

thecynefin.co

Former name: Cognitive Edge
Reach: More than 2,000 clients and more than 3,000 SenseMaker users
Latest: October 2026 announcement of a Cynefin Field Guide series to mark 27 years of the Cynefin framework

SenseMaker asks people to share short accounts of their experiences and then interpret those accounts themselves, rather than leaving interpretation to an analyst. Because participants classify their own stories, patterns can be seen across many contributions without outside categories being imposed. The results combine qualitative stories with quantitative data.

The firm now trades as The Cynefin Company and The Cynefin Centre, while its legal entities keep the Cognitive Edge name. Its programs cover education, health and care, climate, international development and citizen engagement.


CrowdSmart Labs

Software that structures deliberation between people and AI agents and turns it into forecasts with confidence scores, ranked reasoning and an audit trail.

crowdsmart.ai

Products: Collective Reasoning Foresight, Knowledge Agent and Collective IQ
Deployments: NATO, where AI agents and senior commanders deliberated more than 3,000 ideas across 31 countries (company-reported)
Latest: In April 2026, an ensemble of its Knowledge Agents correctly forecast FDA approval of the obesity drug orforglipron (company-reported)

CrowdSmart started as a platform for predicting startup investment outcomes from the structured judgments of investors. It has since repositioned as CrowdSmart Labs, a “trust layer” for human and AI collective intelligence. In a session, people and AI agents contribute judgments, explain their reasoning and rate one another’s ideas. Bayesian belief networks combine these into a collective forecast that preserves dissent rather than averaging it away.

The company reports that its approach supported 71% portfolio survival for the Titan Fund over an eight-year deployment, against an industry baseline of 38%. It holds four issued US patents. Most of its performance evidence is self-reported.


Originally written by Joseph Mapue in 2021. Fully revised in November 2026.