×
Mantic
AI forecasting system that beat every human in the Metaculus Cup
Last verified September 19, 2026 · Updated daily
What Mantic is building
Mantic builds an AI system for judgmental forecasting: the kind of question where there is no statistical model to fall back on, only reasoning about the world. Will a central bank cut? Will a regulator approve a licence? Will a conflict escalate? The product has three primitives: Horizon Scan (surface the questions that matter to you before they resolve), Forecast (a probability with written reasoning and references), and Monitor (the probability updates as new information lands). Horizons run from one week to one year, the system works from public information, and customers can plug in proprietary data. The technical bet is not a new foundation model. Mantic specialises frontier models for forecasting by running them against historical questions, grading the output, and iterating on scaffolding, tools for expressing probability distributions, and research agents. The careers page describes research projects that last about a week, with internal benchmarks and failure analysis on bad predictions. It is an eval-driven shop, which is what you would expect from a CEO who co-led dangerous-capability evaluations on Gemini. The proof is public and scored. In the Metaculus Cup, Mantic went 4th of 539 humans (Summer 2025), 8th (Fall 2025), 1st (Spring 2026), and 1st again (Summer 2026), where it beat every human and all bots except one called laertes. The site shows the shape of the edge: it fell to 5% on Revolut getting its UK banking licence while the community sat at 52% (restrictions remained); it put 20% on a new BRICS invitation versus the community's 70% (no invitations came).
Why this matters
Forecasting is one of the few AI capabilities you cannot fake: every question resolves, every forecast gets a score, and the leaderboard is run by a third party. That makes Mantic's superhuman claim unusually checkable, and it is why Radical's Aaron Rosenberg could say on the record that companies and government agencies have already integrated the system, with hedge funds and trading firms the keenest. His line: “If Mantic is superhuman as it is, they can make money off of that immediately.” The timing argument is about who else is in the round. M12, Balderton, and Thinking Machines Lab (a frontier lab investing in an applied startup) backing a $25M seed for a 2024-founded company says the market believes the moat is in the harness and the evaluation loop, not in the base model. If that is right, the same playbook (specialise a frontier model, grade it relentlessly on a domain with clean ground truth) is what the next wave of applied-AI companies will run, and Mantic is the reference case.
Open roles at Mantic
3 positions we're tracking. Roles are re-checked daily and removed when filled.
AI Engineer / Researcher, LLM agents (Member of Technical Staff; Camden, London; in-person)
First seen today
Research engineers with strong statistics (implied by the raise and the JD's focus on benchmarks and calibration)
First seen today
Forward-deployed / customer-facing technical roles for finance and government clients (inferred from Rosenberg's comments on integrations)
First seen today
Know when Mantic is hiring before anyone else
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $5.99/week, cancel any time.
Watching Mantic
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+Hiring outlook
One live role on their Breezy board (AI engineer / researcher, LLM agent focus, Camden, in-person five days a week, visa sponsorship for exceptional candidates). No published headcount target, but a $25M seed for a research team that was ~$4M-funded a year ago implies several more technical hires; the team page lists a mix of DeepMind, Citadel, Palantir, Goldman, McKinsey and Bank of England backgrounds, so expect both research and quant-adjacent roles.
Working at Mantic
Mantic is a AI company based in London. Working at a AI company at this stage means opportunity to shape your role based on the company stage.
London is where most Mantic positions are located.
How to actually get hired at Mantic
Why applying the normal way doesn't work
There's an ATS, but at Mantic, referrals get priority. Your cold application competes with sourced candidates and internal recommendations.
Who to contact at Mantic
What to show them
Pick 20 resolved Metaculus questions from the last quarter. Build a small agent (retrieval plus reasoning plus an explicit probability output) and score it with Brier against the community median at the same timestamps. Ship the repo, a one-page failure analysis of the three worst misses, and one scaffolding change that measurably improved calibration. That is literally the week-long project shape their JD describes.
A cold email that works at Mantic
What Mantic screens for
Radical led and put a partner on the board; M12 and Balderton co-invested; Thinking Machines Lab took a position. That combination says the investors see a distribution business (finance and government buyers who already integrate the system, per Rosenberg) rather than a research bet. The hiring window is now: the live role is in-person in Camden, projects run about a week, and the JD asks for LLM-agent scaffolding plus real statistics. A team this small hiring one researcher at a time means a specific, scored piece of work will get read.
Customize your CV for the Mantic role. Matching the job description language helps clear ATS filters.
Don't make these mistakes
Do not open with prediction-market enthusiasm or 'AI will predict everything'. Toby wrote an Oxford PhD on LLM governance and misuse and co-led dual-use capability evaluations inside Gemini; he has spent years on exactly how these systems fail, so hand-waving about superhuman AI is noise. Ben's PhD is in meta-learning and graph neural networks and he ran research at a company optimising steel production; do not pitch him a bigger model. Lead with calibration: a concrete forecasting question you worked, how you scored yourself, and where your Brier score went wrong.
Mistakes that kill Mantic applications
Generic CVs stand out at a -person company — and not in a good way. Fastest path to rejection.
Don't open with what you want. Open with what Mantic is dealing with right now — Forecasting is one of the few AI capabilities you cannot fake: every question resolves, every forecast gets a score, and the leaderboard is run by a third party, and what you'd do about it.
Most AI applications get ghosted. A day-five follow-up can double your response rate.
Applying to Mantic? Get the contact, not the form.
The Mantic interview process
4 stages · 14 days typical · take-home: yes · modelled from similar companies
We don't yet have verified candidate reports for Mantic. What follows is the typical process for a -person AI company — treat it as a model, not confirmed detail.
Interview stages
Recruiter Screen
Phone or video · 30 min
Basic qualification and logistics
Recruiter or HR
Hiring Manager Interview
Video call · 45 min
Role fit and experience deep-dive
Hiring manager
Technical/Functional Round
Video call · 60 min
Skills assessment and problem-solving
Team members
Final Round
In-person or video · 60 min
Culture fit and cross-functional alignment
Senior leadership
Mantic take-home assignment
Mantic includes a take-home exercise in their interview process. For AI roles, this typically involves a practical problem that takes 2-4 hours. Focus on clean, working code over premature optimization. They're evaluating how you think and communicate, not just the solution.
Mantic interview timeline
Expect days total. Compared to similar AI companies (14 days median), Mantic is about average.
Interviewed at Mantic?
Tell us how it went — stages, questions, timeline. Takes 90 seconds and it's how this page stays accurate for the next person.
Submit your Mantic interview experience →Mantic jobs, frequently asked questions
How many jobs does Mantic have open?
As of September 2026, Mantic has 3 open positions.
Does Mantic hire remotely?
Mantic doesn't have remote openings at the moment. All roles are in London.
What roles is Mantic hiring for?
Mantic is hiring across Engineering, Finance. The most recent opening is AI Engineer / Researcher, LLM agents (Member of Technical Staff; Camden, London; in-person).
How do I apply for a job at Mantic?
Use the apply links above, or check our guide to getting hired at Mantic.
Does Mantic respond to cold emails?
Response rate data for Mantic not yet confirmed.
Who is the hiring manager at Mantic?
At this size, hiring is usually run by the hiring manager.
How competitive is it to get hired at Mantic?
Expect 100-250 applicants in the first two weeks for AI roles at this size. Apply within 72 hours for best odds.
How many rounds is the Mantic interview?
4 stages: Recruiter Screen, Hiring Manager Interview, Technical/Functional Round, Final Round.
Is the Mantic interview hard?
Expect technical depth and system design, not algorithm trivia. Candidates report Technical Interview as the toughest stage.
Does Mantic give a take-home task?
Yes, Mantic includes a take-home assignment.
How long does Mantic take to get back to you?
Around 14 days across the full process.
What should I prepare for the Mantic interview?
Focus on technical depth and system design. At people, they're testing whether you can operate without process, not whether you memorised algorithms.
Where is Mantic based?
Mantic is headquartered in London, US.
Get Mantic roles before they're posted
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $5.99/week, cancel any time.
Watching Mantic
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+Related
More AI companies hiring
Other startups we track in the same space.