A product-marketing plan for the real-time prediction protocol Somnia is building. Polymarket owns the slow question, will it happen. The open lane is the fast one: what is happening right now, priced and settled while the event is live. This page segments the market, maps the field, and turns the JD into a plan. Working demo in ★. Public data only, figures approximate.
The thesis. Prediction markets are the breakout crypto category of the cycle, and the leaders won it on a slow product. Polymarket settles through an optimistic oracle: hours when clean, days when disputed. Kalshi is a regulated US exchange, off-chain and permissioned. Both resolve later, an election, a Fed decision, a season. Neither is built for the moment the event is actually happening.
Somnia's sub-second finality, six-figure throughput and sub-cent fees open the lane the incumbents structurally cannot serve: live in-play markets that open, trade and settle inside the event, resolved by AI agents in seconds with an onchain receipt. Somnia already ships the proof, Prophecy Social and Vectra. The marketing job is to name that lane and own it first.
The number that frames it: in-play is already ~50% of a $100B+ sportsbook handle and sports is 80-85% of prediction volume, yet the permissionless-and-sub-second quadrant is empty (§03). The biggest pool in the category is served by no one.
The one line I would run: Polymarket is where you bet on the outcome. Somnia is where you trade the moment.
Speed you can feel. Not a benchmark on a slide, a market that resolves while the crowd is still reacting. The product's edge is latency, so the marketing has to make latency legible to a non-technical trader.
Turn a technically ambitious protocol into positioning traders repeat, launches that land, and a qualified builder base. Own what it is, who it is for, why it matters, and what makes it different, end to end.
The right way to read this market is by segment, not by logo. The value has consolidated into two segments the incumbents own, and one segment nobody serves.
| Segment | Who owns it | Settlement need | Served today? |
|---|---|---|---|
| Politics & geopolitics | Polymarket (offshore), Kalshi (US) | Resolve after the event; latency is fine | owned, saturated |
| Resolved-outcome sports (final score) | Kalshi (sports = most of volume), Polymarket US | Settle on official result, minutes | owned, contested |
| Long-tail culture & niche | Polymarket breadth | Cheap creation, flexible resolution | thin, creation-gated |
| Live in-play (next goal, round, candle) | Nobody at sub-second | Open, trade and settle inside the event | unserved ★ |
In-play is the largest and most throttled pool in the category, and it is the one no rail can price safely today. That is the segment this plan goes after.
Each player against the one question that matters here: how fast can it settle a live moment, and what stops it from going faster.
| Player | Venue | Settlement & latency | Owns today | Structural ceiling | Somnia's wedge |
|---|---|---|---|---|---|
| Polygon · USDC · CLOB | UMA optimistic oracle. ~41 min median, days when disputed | Liquidity, breadth, brand; ICE $2B, US re-entry via QCX | Optimistic oracle needs a liveness window plus a human token vote; whale votes have mis-settled $100M+ markets | Sub-second, deterministic AI settlement of objective live events | |
| CFTC DCM · off-chain | Admin resolution on official data; minutes | US legitimacy; ~$178B annualized, sports-led | Off-chain, KYC-gated, US-only, contract-by-contract approval; no permissionless or composable markets | Permissionless, global, onchain, composable, instant listing | |
| Abstract / Linea / BNB | Chainlink CRE oracle + BNB blocks | Editorial distribution (Decrypt, Rug Radio) | Distribution-led; resolution rides oracle cadence, no sub-second in-play | Own real-time depth while it owns media reach | |
| Base · CLOB | Fixed windows on ~2s Base blocks | Base-native orderbook, Coinbase Ventures | Caps at ~5-minute markets; Base not tuned for per-tick | Per-second and per-tick binary markets | |
| EVM multi-chain infra | Bookmaker feed + post-event resolver tx | Liquidity infra behind many betting frontends | Bookmaker-style, peer-to-pool caps; payout waits minutes to hours | Peer-to-peer price discovery settled in-block | |
| Solana | Committee oracle at expiry; can haircut winners | Real-time matching on Solana speed | Fast matching, discrete resolution; not EVM; winner haircuts | EVM familiarity, with settlement in-block rather than matching alone | |
| Arbitrum Orbit · Chainlink | 250ms blocks, but payout minutes-to-hours | Onchain sports & live betting rails | Fast blocks, slow settlement; fixed risk caps on live | Collapse matching and settlement into one moment | |
| UMA Truth Bot, Sportstensor, World | AI proposes, but a dispute window or Chainlink average still gates it | The emerging AI-resolution narrative | Every one separates the fast layer from settlement; none is truly sub-second end to end | End-to-end trustless in-block AI resolution on a sub-second L1 |
Regulation cleaved the category in two. One lane is a KYC-gated, off-chain US stack (Kalshi, Polymarket US) that ICE validated with a $2B commitment and that surrendered every onchain property to get there: permissionless access, crypto rails, instant listing, global reach. The other lane is permissionless and global, open to the 100+ countries and the markets too niche for a US exchange. The institutional money built the walled lane. The open global rail underneath it is unbuilt, and that is the opening.
Figures are public estimates and move fast; sources in §10. Named target partners are held in my application materials.
Plot the field on the two axes that matter: how fast a market settles, and whether anyone can list and trade it without a gatekeeper. Every player is fast-but-permissioned or permissionless-but-slow. The top-right is empty.
Positions are directional, from public settlement mechanics (Polymarket ~41 min median and days when disputed; SX 250ms blocks but minutes-to-hours payout; Limitless ~5-minute windows on Base). The read is the geometry: the permissionless-and-sub-second quadrant belongs to Somnia alone.
Polymarket and Kalshi run a deliberate delay on live sports because their rails get picked off by latency arbitrage. They are advertising the exact product they cannot ship. Sub-second settlement turns that defensive delay into a category.
Every rival separates matching from settlement: Solana matches in 400ms then waits on a committee, SX runs 250ms blocks then settles in hours, UMA prices instantly then opens a dispute window. These are architectural ceilings, not tuning problems.
Somnia's move is to collapse pricing, matching and AI resolution into one in-block moment. That is the one thing none of the incumbents can retrofit onto their chain, and it is what makes per-second markets offerable.
AI settlement is the wedge and the exposure. Production proof exists (UMA's Optimistic Truth Bot settles most markets within hours at ~90% accuracy; Sportstensor and others price live), but all of it keeps a human or token-vote backstop, and UMA's own whale-vote failures show automated resolution is contested on subjective calls. Somnia's differentiation is end-to-end trustless in-block resolution. Marketing has to lead with objective, machine-checkable moments (a goal, a candle, a round) where the receipt is unarguable, and treat subjective markets as a later, carefully-scoped step. Credibility is the scarce asset, so we do not overclaim.
Solana already matches sub-second and has distribution. A Drift or Velocity relaunch that adds deterministic in-block resolution could contest the real-time claim before Somnia builds liquidity. EVM finality alone is not a moat.
Polymarket or Kalshi could shorten resolution or layer AI proposers. With $20-30B a month and locked distribution, a good-enough live product could box out a new L1. The counter is speed to own the narrative now.
The thesis rests on trustless AI adjudication. A single high-value mis-settlement of a subjective moment hands the safe-resolution narrative back to Kalshi and UMA. Scope, transparency and the public receipt are the mitigation.
The JD names three audiences: traders, builders, partners. Each needs a different message and a different funnel. One narrative spine, three translations.
Want action, speed and edge. The message is the live wire: markets that move and settle in seconds. The funnel is a KOL clip, to a live event market, to a first trade, to a habit. Micro-stakes lower the first-trade barrier.
Want a reason to build markets and apps here rather than on Base or Solana. The message is the primitives: sub-second settlement, AI resolution, cheap market creation, EVM tooling they already know. The funnel is docs, to a template, to a live market.
Streamers, esports orgs, sports media, crypto influencers. The message is a new format: a live market their audience trades during the stream. The funnel is a co-marketed launch moment with a ready-made KOL kit.
The audience that already lives on speed and novelty. Real-time crypto and culture markets are native to them. This is the wedge audience for the first wave of volume.
The largest prize (sports is most of Kalshi's revenue) and the clearest in-play use case: next round, next map, next elimination. Reached through orgs and streamers, not crypto channels.
Delphi, Messari, crypto media, the Prediction Conference circuit. The message is the category thesis, backed by product proof. Partner with comms so the thesis shows up in third-party writing.
Sequence: win the degen and crypto-native trader first (fastest to activate), use that volume as proof to bring builders and KOLs, then expand into sports and esports where the in-play format is largest.
A marketing plan should be demonstrable. I built a working prototype of the exact product moment the wedge is about: live markets with odds moving in real time, then AI-agent settlement in seconds with an onchain-style receipt. It runs in the browser, no wallet, no chain. It is a marketing prop I built for this application, and it is not Somnia code.
Odds that move per second, a market that resolves inside the event, and settlement you can read. This is what "real-time" has to feel like in every asset we ship.
The same widget, embeddable, becomes a landing-page hero, a KOL demo, a conference booth screen and an X card. One prop, many launch surfaces.
Simulated prices and a mock receipt, modelled on Prophecy Social's real AI-agent settlement. Built by me for this application. Not production code, and never presented as Somnia's.
The JD asks the PMM to own what it is, who it is for, why it matters and what makes it different, end to end. Here is that messaging house, drafted.
| Layer | The message |
|---|---|
| One-liner | Trade the moment. Real-time prediction markets that settle while the event is still happening. |
| What it is | A prediction protocol on Somnia where markets open, trade and resolve in seconds, settled by AI agents with an onchain receipt. |
| Who it is for | Traders who want the live wire, builders who want fast market primitives, and partners who want a new format their audience can play during a stream. |
| Why it matters | The most-watched moments (a match, a round, a candle) had no live market fast enough to trade them. Now they do. |
| What makes it different | Sub-second settlement, sub-cent fees and AI resolution let us serve in-play and per-second markets the settlement-heavy incumbents structurally cannot. |
Odds that move in real time and settle in seconds. Proven through the live demo, KOL clips and event streams, not a benchmark chart.
AI agents settle with a public receipt: sources, findings, per-agent votes. Fast and auditable, against multi-day oracle disputes.
Near-zero creation cost makes the long tail viable. A market of markets, permissionless, created by anyone, on any live moment.
Guardrail: the chain is early, so the claims stay concrete and demonstrable. We market what the product does today (Vectra, Prophecy Social) and frame roadmap items as roadmap. Credibility is the scarce asset in a category full of hype.
Every responsibility from the posting, mapped to a concrete move.
| JD responsibility | My plan |
|---|---|
| Own positioning and messaging end to end | Ship the messaging house in §05 as a living doc: one-liner, pillars, per-audience translations, and a claims register that separates live from roadmap. Everyone briefs from one source. |
| Turn complex product strengths into credible narratives | Convert latency and AI resolution into things you can feel: the live demo in ★, event-stream markets, resolution-receipt explainers. Show the speed, do not assert it. |
| Lead go-to-market for launches and incentive moments | Run the launch playbook in §07: a repeatable kit for each market launch, partner activation and incentive campaign, with owners, assets and a measurement plan per launch. |
| Define and run the marketing roadmap with product and community | A quarterly roadmap that sequences launches so each builds on the last: degen crypto markets, then builder tooling, then sports and esports. Standing sync with product and community leads. |
| Grow the qualified builder base with always-on programmes | A builder funnel: docs and a market template, a live-market quickstart, a market-creation bounty, and a recurring builder call. Track builders to first live market, not signups. |
| Run trader, market and KOL research; feed it back to product | A standing loop: talk to traders and KOLs, watch what markets they actually trade, and feed structured findings into product, pricing and creative each cycle. The voice-of-market I ran as an analyst and in CS. |
| Brief and create content that earns attention in Web3 | Launch threads, research-led posts, explainer assets, dashboards, KOL kits and conference material. A weekly cadence anchored to the category thesis, not random posting. |
| Partner with comms on press, podcasts and thought leadership | One narrative spine feeding comms so the category thesis shows up in Delphi, Messari, media and the Prediction Conference circuit, backed by product proof. |
Launches are the core of this role. One repeatable structure, run for every market launch, partner activation and incentive moment, so the team gets faster each time.
| Phase | Moves | Assets |
|---|---|---|
| Pre-launch | Lock the narrative and the one metric that defines success; line up KOLs and partners; seed the thesis with research and press under embargo | Messaging brief, KOL kit, embargoed research post, teaser demo |
| Launch moment | Tie the launch to a live event where in-play is obviously better; go live across X, streams and partner channels at once | Launch thread, live demo, stream overlay, explainer video, incentive campaign |
| Sustain | Daily market callouts, KOL trades, resolution-receipt clips that build the speed-and-trust story; recap the numbers | Data cadence, receipt clips, weekly recap, leaderboard |
| Learn | Post-launch readout against the one metric; structured feedback to product; fold wins into the next launch kit | Readout doc, VoM findings, updated playbook |
Partner with an esports org for a live event where the audience trades next-round and next-map markets during the stream. The format is the ad.
Real-time markets on the next BTC candle, resolved in seconds. Native to the degen audience, easy to clip, a direct jab at slow settlement.
A creation campaign: anyone spins up a market on any live moment in minutes. Showcases the market-of-markets thesis and pulls in builders.
Distribution note: web3-native audiences do not come from search. They come from X, Telegram, KOL clips, and quest platforms (Galxe, Layer3). The plan is built for that motion, seeded in communities and through creators, with SEO and AI-answer visibility as a secondary layer for the category thesis.
The JD wants someone confident with data. The metrics that matter for a real-time prediction protocol, and the ones I would not chase.
Weekly active traders, new-trader activation, markets traded per user, notional volume. The core health of a trading product.
One defined metric per launch, tracked against target, plus reach, first-trade conversion and cost per activated trader.
Builders to first live market, markets created, and markets that attract real volume. Supply that gets used, not signups.
Share of voice on "real-time" and "in-play" prediction markets across X, research and AI answer engines. Are we the name for the lane.
Return rate after the first market, and after the first resolution. Does the speed create a lasting habit beyond the launch spike.
Impressions with no downstream trade, incentive-farmed volume that leaves, and follower counts. Watched, never the target.
Outside-in, from public data only, no confidential Somnia information. The live demo is a prop I built for this application. Figures are approximate public estimates and move fast in this category; confirm against internal data before use.
Somnia: somnia.network · docs (concepts) · 2026 outlook
Prediction apps: Prophecy Social · Vectra
Capital: $270M ecosystem capital
Ecosystem data: Messari, State of Somnia Q4 2025
Polymarket: UMA resolution docs · ICE $2B · QCEX $112M
Kalshi: overview · valuation & volume
Others: Myriad · Drift BET · Azuro
Measurement tooling: Ahrefs Brand Radar · Dune · onchain analytics
Which specific protocol the JD means by "a prediction protocol built for real-time markets" is something I would confirm with the hiring manager. I have led with Vectra and Prophecy Social as the public evidence. Named launch partners and the full plan are held in my application materials.
Independent homework by Edward Tay for the Somnia Web3 Product Marketing Manager, Prediction Markets application. Public data only, no confidential information.