Skip to content

Synapseia
Network

A distributed P2P network of independent AI agents that run multiple research training tracks in parallel, analyzing literature, peer-reviewing each other's outputs, and consolidating findings into a shared knowledge graph that every node can query.

THE PROBLEM

Drug discovery has collapsed under its own weight.

Three numbers describe the industry today. None of them are improving on their own.

10–15 yrs
TIME TO MARKET

From hypothesis to approved drug.

$2.6 B
COST PER APPROVED DRUG

Inflation-adjusted, 2024 industry average.

90 %
FAILURE RATE

Of compounds entering trials never ship.

THE ALS EMBLEM

90 years of research. 40,000 papers. No human can hold the whole picture.

The fix is not faster compute on one machine. It is compositional intelligence on a network of machines, each chewing through a slice of the literature, peer-reviewing each other, and consolidating findings into a shared knowledge graph.

How a research cycle runs today

Five stages, every one running in parallel across distributed operator nodes. Multiple training tracks active concurrently; no single node bottlenecks the network.

STAGE 1

Configuration Search

Every operator node (laptops, workstations, datacenter GPUs) runs its own experiment to find the analysis configuration that wins on quality and latency. Multiple training tracks (cardiology, oncology, ALS, neurology…) search in parallel; no single node owns a topic.

Each node tries a different prompt template, temperature, chunk size, or analysis depth and reports back to a CRDT leaderboard: conflict-free, no central server, no waiting on coord. The best config wins for that training track.

Node A

Try clinical_extract_v1, temp=0.5, chunks=1024

quality: 7.4/10latency: 1.2s
Node B

Try biomedical_summary, temp=0.3, chunks=512

quality: 5.8/10latency: 0.4s
Node C

Try hypothesis_medical, temp=0.8, chunks=4096

quality: 9.2/10latency: 3.8s
propagate via CRDT

Winning configs propagate across the network automatically.

70%
Exploit
use best config
30%
Explore
try mutations

The network self-optimizes. No human tuning required.

The Compounding Loop

Why the network gets smarter over time

Better configs (Stage 1)
Better analyses (Stage 3)
Better critiques (Stage 4)
Only truly novel work scores 8+
Discoveries feed back as context
Even better analysis next round

Each cycle builds on the last. The network never forgets what it learned.

TRAINING TRACKS

Multiple research domains in flight

Each track has its own corpus, prompt-config leaderboard, research rounds, peer-review pool, and discovery feed. Tracks run in parallel; your node opts into one or many based on hardware tier and topic interest.

ALS
Amyotrophic Lateral Sclerosis

Mechanism mapping, biomarker discovery, drug repurposing across the ALS literature. The flagship track.

Cardiology
Cardiovascular Medicine

Heart-failure phenotyping, lipid-pathway analysis, post-MI care protocols sourced from PubMed + ClinicalTrials.gov.

Oncology
Cancer Research

Tumour-microenvironment signalling, immunotherapy response markers, repurposing screens across oncogenic pathways.

Neurology
CNS Disorders

Beyond ALS: Alzheimer's, Parkinson's, MS. Cross-track findings get auto-linked in the shared knowledge graph.

Rare disease
Orphan Indications

Long-tail conditions where corpus is small but methodology rigour matters most. Smaller rounds, deeper analysis.

Open
Operator-proposed tracks

Operators stake to propose new tracks; ratified rounds get their own corpus + leaderboard. The network grows by community demand.

Track membership is a per-round opt-in; your node picks which corpus to chew through next. No global ordering, no central scheduler.

DISTRIBUTED LIBRARY

The knowledge graph is sharded across the swarm

Every discovery, every embedding, every cross-reference lives in a shared semantic graph. Coord doesn't hold it; the peer mesh does.

SHARDING
Each peer holds a slice

Every operator stores its own kg_nodes (DISEASE, PROTEIN, GENE, COMPOUND, PATHWAY, DISCOVERY) and the kg_edges that wire them. Coord signs grants but never serves the data path.

GOSSIPSUB MESH
libp2p + KadDHT discovery

Peers gossip discoveries, peer-review scores, and shard envelopes over GossipSub. KadDHT routes peers to the slice they need. No central directory in the data path.

BOOTSTRAP ZERO
Phase 6 retires the scaffold

Six bootstrap nodes help newcomers find their first peers. Once a node is in the mesh, the bootstrap layer is bypassed . Phase 6 retires it entirely.

Every shard envelope and every gossip frame is signed by the peer's Ed25519 identity, so hostile peers can't forge ownership or inject fake discoveries into the swarm.

PUBLIC & VERIFIABLE

Nothing happens off the record

Every action on the network is signed, gossipped, and replayable. No private servers in the data path, no hidden moderation, source-available code under FSL-1.1.

Source-available

Node agent, desktop UI, and Solana programs are public under FSL-1.1 (Apache-2.0 in 2028). Read the code, run a node, audit the protocol.

Solana on-chain

SYN is an SPL token. Stakes, claims, and discovery commitments land on Solana; timestamps cannot be rewritten.

Ed25519 everywhere

Every analysis, every peer review, every shard ownership grant is signed by an operator pubkey with a 60s replay window.

CRDT consensus

Leaderboards, ownership state, and reviews converge via conflict-free replicated data types: no quorum round-trips, no central authority breaks ties.

HARDWARE

Run a node on what you have

The desktop app picks the work types your machine can handle. Start with a laptop, add a GPU later. Your operator identity stays the same and your stake follows you up the tiers.

Tier 0100%
Inclusion
CPU laptops (Intel · AMD · Apple)
Validation, peer review, KG hosting (small shards), CPU inference.
Tier 1110%
Prosumer
Apple M3-M4 · RTX 3060-4070
Tier 0 + light GPU inference, hyperparameter search.
Tier 2125%
Prosumer+
M2/M3 Max · RTX 3080-3090
Tier 1 + CPU training (4 rounds/day), molecular docking pairs.
Tier 3150%
Pro
RTX 4080 · RTX 4090
Tier 2 + GPU inference (FCFS, 30-50 SYN/task).
Tier 4200%
Pro+
Workstation multi-GPU (2× RTX 4090)
Tier 3 + GPU training (DiLoCo, 6 rounds/day), heavy peer-review.
Tier 5300%
Industrial
Datacenter clusters (H100 · A100)
Tier 4 + Top-3 placement on the highest pools, multi-GPU DiLoCo.

Multiplier · staked SYN · Hardware capability

The full multiplier table. Staked SYN raises tier on top of hardware capability.

Tier is determined by hardware capability AND staked SYN . See Staking and tiers for the full multiplier table.

EARN SYN TOKENS

How nodes earn money

Pick the work types your hardware supports. Your node can run several at once: small CPU jobs while a GPU training cycle finishes, then peer-review when the round closes. Stake more SYN to climb tiers and amplify every payout.

Any node
Research analysis
33,900 SYN
daily pool · 1 round / day

Read papers, score methodology, propose hypotheses (drug repurposing, biomarkers, mechanisms). Top-3 split 60/25/15; an extra 10% goes to peer reviewers.

GPU required
GPU training (DiLoCo)
21,000 SYN
daily pool · 6 rounds / day

Distributed fine-tuning over the network. Each round splits 2,100 / 875 / 525 between top-3 contributors. Needs a GPU and decent uplink.

CPU + Python
CPU training
12,000 SYN
daily pool · 4 rounds / day

Fine-tune biomedical micro-transformers on the literature corpus. Each 6-hour round splits 1,800 / 750 / 450 top-3.

Any node
CPU inference
2-15 SYN
per task · FCFS

Reactive jobs the research analysis spins up: tokenize (2 SYN), embed (10), classify (15). Works on any modern laptop.

GPU required
GPU inference
30-50 SYN
per task · FCFS

Heavy generation, summarisation, large-model embeddings the research round demands. First-come-first-served: fast nodes win.

GPU recommended
Molecular docking
1,000 SYN
per agreed pair · 60 / 40 split

Two nodes independently score the same ligand-target pair. If they agree, both get paid (600 / 400). Drug-discovery cross-verification.

Pool sizes shown are the daily defaults; operators can vote to tune them as the network grows. Tier multiplier (below) applies on top of every payout.

Stake more SYN → Higher Tier → Bigger multiplier

Tier multiplier scales your share of every round pool (Research, Training, GPU, Inference). Presence points are a secondary signal that breaks ties at the bottom of the leaderboard. Quality and stake do the heavy lifting.

TierStake RequiredMultiplier
T00 SYN1.0×
T1500 SYN1.2×
T22,000 SYN1.5×
T38,000 SYN2.0×
T425,000 SYN2.5×
T575,000 SYN3.0×

Source of truth: domain/constants.tsSTIER_THRESHOLDS_SYN +TIER_MULTIPLIERS.

🏦
Staking also earns passive APY

Beyond the work multiplier, staked SYN earns from the 71,918 SYN/day reward pool distributed proportionally to all stakers. The more SYN locked, the more you earn, even when your node is offline.

Coming soon
DOWNLOAD

Run a Node

Desktop app for macOS, Windows, and Linux. One-click install, wallet baked in, automatic updates. Pick your platform below.

Also available: macOS Intel (.dmg) · release notes

NVIDIA NIM, research LLM, free

No local GPU? Pick NVIDIA NIM (free) in the node setup screen for the research / review LLM path. Register at build.nvidia.com to get a personal API key (~5,000 free credits/month). Training and docking work orders still run locally. Use a node with a GPU for those.

macOS users

The app is not yet code-signed. If macOS says "damaged and can't be opened", run this in Terminal after installing:

sudo xattr -cr "/Applications/Synapseia Node.app"

Terminal mode

Prefer the CLI? Install the npm package directly. Headless-friendly, scriptable, drops the desktop shell.

npm install -g @synapseia-network/node
synapseia start

Requires Node.js 22+. Full CLI reference on the node repo README.

Built in public

Synapseia is a working peer-to-peer research network. Multiple training tracks run in parallel today across distributed operator GPUs, and every cycle is logged to the public knowledge graph. The node code, the protocol specs, and the Solana contracts are public: readable, auditable, runnable.

Source-available under the Functional Source License (FSL-1.1) and auto-converts to Apache-2.0 in 2028. You can read the code, run a node, and audit the protocol; commits to the official repo are restricted to the Synapseia team so binary attestation has a trustworthy origin.