Home/AI/@cludeproject
62
Score · promising
Scored 1w ago

@cludeproject

CludeAI

CludeAI is building a cognitive memory engine for AI agents with typed memory structures, differential decay rates, and graph-based recall. The project has a working SDK on GitHub, claims 1.96% hallucination vs 15.2% for competitors on HaluMem benchmark, and has received grants from Google Singapore ($25K) and Solana Foundation ($10K). They've launched a token ($CLUDE) on pump.fun and are training what they claim is the world's first memory-specific language model (4B parameters on 4.9M examples). The technical approach is differentiated with five memory types (episodic, semantic, procedural, self-model, introspective) each having distinct decay curves.

Token

Current price
$0.0007260
3h ago
Liquidity $135kVol 24h $252kPool since 2026-02-24

Where to trade

Links go to third-party venues. Always verify the contract address. Not financial advice.

Contract & market health

Mixed50% · 3/4 contract checkschecked 4d ago
Contract — one ❌ is a red flag
  1. Contract can no longer mint tokens?mint authority revoked
  2. Nobody can freeze, blacklist or pause?freeze authority revoked, no transfer hook
  3. Sellable, without abusive tax?standard SPL token, no close/balance authority
  4. Liquidity out of the team's reach?Solana LP lock: check manually (LP burn / locker)
Distribution, market, signal — 10% each (◐ = 5%)
  1. Top 10 wallets (excl. pool/staking/locker) ≤ 30%?49.6% (goplus)
  2. No single wallet (excl. pool) above 10%?largest: 11.2% (wallet 7udbgT6w…)
  3. ≥ 500 holders, growing over 7d?3814 holders (7d growth: not tracked yet)
  4. Team / insider allocation vested on-chain?no locker/vesting identified — manual check (docs, transparency page)
  5. Pool ≥ $50k?liquidity $106k
  6. 24h volume / liquidity between 0.2 and 5?vol $8k / liq = 0.1 (dormant token) · 45 buys / 77 sells
  7. No overhang (mcap ≥ 30% of FDV, FDV ≤ $10M)?FDV $472k, mcap $472k (100%)
  8. Healthy timing (pool ≥ 7d, no +300% pump, no wipeout)?pool 207d, +10%/24h, +1%/6h, 67% of known high
  9. X account alive (≥ 3 months) and shipping (≥ 3 tweets/7d)?account 195d old · 0 tweets/7d
  10. Real product and ≥ 2 scouts agree?5 AI green flags · 1 scout

Sources: GoPlus (contract), the chain explorer (holders), DexScreener (market), ohmybird (signal). ❓ means we couldn't verify it automatically — check it yourself. Not financial advice.

AI Analysispromising

Confidence
72%

CludeAI is building a cognitive memory engine for AI agents with typed memory structures, differential decay rates, and graph-based recall.

The project has a working SDK on GitHub, claims 1.96% hallucination vs 15.2% for competitors on HaluMem benchmark, and has received grants from Google Singapore ($25K) and Solana Foundation ($10K).

They've launched a token ($CLUDE) on pump.fun and are training what they claim is the world's first memory-specific language model (4B parameters on 4.9M examples).

The technical approach is differentiated with five memory types (episodic, semantic, procedural, self-model, introspective) each having distinct decay curves.

Green flags: Real working product with SDK on GitHub and npm package installable today · Concrete technical differentiation: typed memory with differential decay, entity graphs, and lateral recall system · Legitimate institutional backing: $25K Google Singapore grant, $10K Solana Foundation grant, mentioned pump.fun backing · Measurable performance claims with specific benchmark (1.96% hallucination on HaluMem vs 15.2% next best) · Small following (2,388) with recent account (March 2026), genuine early-stage signal

Red flags: Dual enterprise/crypto strategy may indicate unclear business model focus · Memory model training claims (4B params, world's first) are ambitious and unverified · Token launched on pump.fun (meme launch platform) for infrastructure product raises questions about tokenomics fit · Limited evidence of real adoption or enterprise customers despite enterprise claims

Token
$Clude
Chain
Solana
Stage
mainnet+live
Category
agent memory infrastructure

Signal Timeline

WA
AFirst discovered·1w ago

Score breakdown0–100

🎯Scout conviction
+17.5 / 35
📚Scout consensus
0 / 10
🪪Profile & earliness
+17 / 20
✍️Substance
+16 / 20
🤖AI verdict
+21.3 / 30
⚠️Penalties
-10 / 40
62
Below threshold (74)
Watching for additional signals.
Followers
2.4K
Account age
6mo
Scouts
1
First seen
1w ago