Versus · Mantic

The instrument and the oracle.

Mantic is genuinely impressive — a DeepMind/Oxford team that competes at the top of the Metaculus Cup with an RL-trained engine backed by $4M in pre-seed funding. They answer discrete questions with calibrated, tournament-tested accuracy. This page is not a hit piece. It is an honest read of when you need Mantic, when you need Palanor, and why most stewards want both.

The core distinction

Mantic predicts events.

The unit is a question: “Will the Fed cut before Q2?” “Will this CEO depart by year-end?” A question is discrete, resolvable, and scoreable. Mantic trains against 10,000+ resolved questions using reinforcement learning and a backtesting engine that evaluates predictions in milliseconds. The output is a probability with a reasoning chain and references. It is an oracle — you ask, it answers, the answer lands on a public track record.

Palanor instruments the watching.

The unit is a signal: oil price, Fed Funds, AI capex, discourse sentiment,4,600+ of them moving every day. A signal does not resolve — it moves. Palanor tracks the lattice continuously, weaves signals into scenarios, interprets them through schemas, and reads the whole through Numen. The output is not a probability on a question; it is an understanding of what the world is doing and what it means for your stewardship. It is an instrument — you watch through it.

A PE firm might use Mantic to answer “Will this portco’s CEO depart before Q1?” and Palanor to continuously monitor the signals that would tell them the CEO is in trouble before the question is ever asked.

The two companies

Palanor, Inc.

Founded: 2026 · Delaware C-corp
Entity: Palanor, Inc.
Funding: Bootstrapped · pre-seed target $750K–$1.5M after pilot validation
Team: Tim Woodring (CEO/CSO, PhD) · Claude (product engineer + brand custodian + system operator)
Pricing: Transparent tiers — Reader $0 · Creator $20/mo · Researcher $80/mo · Watcher $4,500/mo · Steward $10,000/mo · Council $35,000/mo
Proof: Public Accuracy Ledger — every claim Brier-scored in public, misses included

Mantic Technologies Ltd

Founded: 2024 · London
Entity: Mantic Technologies LTD
Funding: $4M pre-seed (Episode 1, DRW, DeepMind/Anthropic angels)
Team: Toby Shevlance (CEO, ex-DeepMind, Oxford PhD) · Ben Day (CTO, Cambridge PhD) · team backgrounds at Google, Citadel, Palantir, Goldman Sachs, McKinsey, Bank of England
Pricing: Sales-led · no public pricing · "book a demo"
Proof: Top Metaculus Cup placements (4th of 539 humans Summer 2025; 1st in 2026 rounds)

Axis by axis

Dimension	What it means	Palanor	Mantic	Verdict
Core unit	What the product fundamentally works on.	● Signals — 4,600+ continuous series tracked daily, woven into scenarios, indices, and schemas	● Questions — discrete, resolvable, binary/multiple-choice outcomes on a known horizon	Different products. Palanor tracks what the world is doing; Mantic predicts whether a specific thing will happen by a specific date.
Forecasting methodology	How the prediction is actually produced.	◐ Stake engine — 5-model ensemble over lattice evidence, prediction-market priors, Brier-scored calibration. Palanor Foresight extends signals with multivariate projection.	● RL-trained against 10,000+ resolved questions, backtesting engine, autonomous research pipeline. Tournament-validated.	Mantic wins on event forecasting — that is their entire product. Palanor wins on signal interpretation — that is ours. The methodologies are complementary, not competing.
Continuous monitoring	Watching the world around the clock, not just when a question is asked.	● The lattice ticks daily. 4,600+ signals, 40 Custom Indices, 17 Schemas, 13 Currents — all on cron. Council agents publish continuously.	◐ Horizon Scan + Monitor surface emerging issues and update predictions, but the core unit is the question, not the signal stream.	Palanor. The watching is the product. Mantic watches in service of the next prediction; Palanor watches because the watching is the point.
Scenario modeling	Weaving signals into named futures with probability, impact, and leading action.	● Four-board scenario pages · Lattice Canvas · Futures Frames · three creation paths (Guided · Library · Lattice). Scenarios hold a thesis with signals, narrative, timeline.	— No scenario surface. Predictions are standalone questions, not woven into org strategy or tested against a risk surface.	Palanor. A prediction without a scenario is a number. A scenario tells you what the number means for your business and what to do about it.
Signal relationships	How signals relate to each other — correlations, weighted edges, lattice topology.	● Correlation engine (3,105 pairs scored daily) · Weave (weighted edges drive scenario math) · Lattice Canvas (typed edges: leads-to, inverse, amplifies, dampens, contingent-on)	— Each forecast is independent. No correlation surface, no relationship graph, no lattice topology.	Palanor. The lattice is the moat. Mantic predicts a question; Palanor tells you which other signals move with it and why.
Custom Indices	Proprietary composite indicators authored by the platform.	● 40 Custom Indices — The Palanor Index, Capital Tightness, AI Margin Compression, Discourse Index, Hesitation Index. Each with methodology page, trademarked, filed Class 036 + 042.	— No composite indicator surface. The output is a probability on a question, not a named, narrated instrument.	Palanor. Custom Indices are the brand asset — named things stewards point at in meetings. Mantic has no equivalent.
Interpretive layer	How the product tells you what the data means.	● Numen — one agent, three surfaces. Personalization rows on every intelligence component, tuned to the user role + disposition. Schemas (interpretive lenses) read signals into stages.	◐ Each prediction includes reasoning quotes and references. No persistent agent, no personalization, no schema layer.	Palanor. Mantic explains a prediction; Numen reads the whole lattice through your specific stewardship.
Tournament track record	Public, calibrated proof of forecasting accuracy.	◐ Accuracy Ledger — every claim Brier-scored in public. Building toward tournament play. The ledger is honest about misses.	● Top Metaculus Cup placements — 4th of 539 humans (Summer 2025), 1st in 2026 rounds. Beat GPT-5.2 High (#178) and Claude Opus 4.5 High (#156). This is their proof point and it is real.	Mantic — today. They have the tournament results. Palanor has the Accuracy Ledger and is building toward the same proof. Concede the gap honestly.
Org calibration	Tuning the product to read the world through your specific business.	● Business profile drives Context News ranking, scenario library curation, Schema defaults, Numen framing. The cold-start problem is solved at onboarding.	◐ Enterprise clients can integrate proprietary data sources. The general product is domain-agnostic, not org-calibrated.	Palanor. The business profile is the data substrate that powers everything downstream. Mantic calibrates to your data; Palanor calibrates to your strategy.
Pricing transparency	Can a steward see what it costs before talking to sales?	● Every tier published. Reader free. Researcher $80/mo. Watcher $4,500/mo. Council $35,000/mo. No "book a demo" to learn the price.	— No public pricing. Sales-led, "book a demo," quote-based. Typical of enterprise forecasting at their deal size.	Palanor. Transparency is a brand principle, not a sales tactic. A steward should know what the instrument costs before they pick it up.
API + developer surface	Can developers and other agents call the platform programmatically?	● REST API at api.palanor.com · Palanor MCP (read-only lattice over Model Context Protocol) · developer portal at developers.palanor.com · API tokens, metered tiers.	— No public API or developer surface mentioned. The product is a surfaced application, not a callable intelligence layer.	Palanor. The MCP is the developer wedge — any agent can call the lattice. Mantic is not built to be called; it is built to be asked.
The long horizon	How far forward the product helps you see.	● Signal Watch (continuous) · Scenarios (quarterly + weekly) · Seer Backplan (5-year speculative) · Macro View (daily). The horizon stretches from today to five years.	◐ 1 week to 1 year. Explicitly medium-term. Does not attempt long-horizon scenario backplanning.	Palanor. Mantic owns the 1-week-to-1-year question. Palanor owns the continuous watch plus the 5-year backplan. Different horizons, different jobs.

● full · ◐ partial · — not offered. Mantic facts sourced from mantic.com + public reporting, Sep 2026. Corrections welcome: hello@palanor.com.

Where Mantic wins

Event forecasting. If your core need is a calibrated probability on a discrete, resolvable question — “Will X happen before Y date?” — Mantic is the better tool today. Their RL-trained engine, backtesting infrastructure, and tournament track record (top Metaculus Cup placements, including 4th of 539 humans in Summer 2025) are real and impressive. They raised $4M to build exactly this. We do not pretend to match their tournament accuracy on event prediction — that is their entire product and they are genuine specialists.

If that is your job — a trader pricing event risk, a researcher forecasting election outcomes, a desk that needs tournament-calibrated probabilities on named questions — use Mantic. We mean that. The competitive set is not zero-sum.

Where Palanor wins

Continuous watching.

4,600+ signals on cron. 40 Custom Indices. 17 Schemas. 13 Currents. The lattice never closes. Mantic watches in service of the next prediction; Palanor watches because the watching is the product.

The lattice.

Signals relate to each other. The correlation engine scores 3,105 pairs daily. The Weave lets stewards weight edges that drive scenario math. Mantic predicts a question in isolation; Palanor tells you which other signals move with it and why.

Scenario modeling.

A prediction without a scenario is a number. Palanor weaves signals into four-board scenarios with probability, impact, timeline, and leading action. Mantic has no scenario surface — the prediction is the deliverable.

Numen reads it for you.

One agent, three surfaces, personalization rows tuned to your role and disposition. Mantic explains a prediction; Numen reads the whole lattice through your specific stewardship — CFO sees cost implications, CHRO sees talent dynamics, CEO sees strategic positioning.

Where Palanor answers the same need differently

Palanor Foresight™

Palanor does not predict discrete events the way Mantic does. But every signal in the lattice can be extended forward with three forecast modes — and the third is where the lattice becomes a forecasting engine in its own right.

Mode 1

Continuation

If this signal keeps moving at its current clip, here’s where it lands in 30 days. Linear extrapolation from recent velocity. Honest about being a dumb projection.

Mode 2

Historical Pattern

Looking at past rises and falls and cycles, here’s what the pattern suggests. Ornstein-Uhlenbeck mean-reversion modeling + seasonal decomposition on the signal’s own history.

Mode 3 · Palanor Foresight™

The lattice read

Palanor’s own model — factoring in correlated signals, lattice topology, and Numen’s semantic read of what’s driving the movement. Not just where has this signal been, but where is it going given what the lattice is doing.

Foresight extends every signal, every Custom Index, and every Studies view — on the Terminal Library and the Enterprise Intelligence Platform. It is not a prediction oracle. It is the lattice, projected forward.

The verdict

Use Mantic when

· You need a calibrated probability on a discrete, resolvable question
· Your job is event-risk pricing, election forecasting, or tournament play
· You want tournament-validated accuracy and a public track record on questions
· Your horizon is 1 week to 1 year

Use Palanor when

· You need to watch the world continuously, not just answer a question
· You need signals woven into scenarios with probability, impact, and leading action
· You want to understand how signals relate, not just where each one lands
· You need the reading calibrated to your business, not a generic lens
· Your horizon is today to five years, continuously

Most stewards want both. The oracle answers the question; the instrument watches the world around it.

See the lattice for yourself. Every signal, every index, every scenario — on the Accuracy Ledger.

Our track record See pricing Full comparison →