Subnets / SN22
Desearch SN22
Desearch (subnet 22) is a Bittensor subnet held by 2,004 unique coldkeys. Its liquidity pool holds 7.7k τ, with the alpha token trading at 0.0032 τ. Roughly 4% of tracked holders are in profit — most bought in above today's price. Seller exhaustion is elevated: 90% of recent sellers have emptied their wallets, thinning the supply overhang. Full insider flow, whale tracking and the complete proprietary factor set for Desearch are available to SubnetStats subscribers.
ALPHA PRICE
0.0032 τ
IN TAO
HOLDERS
2,004
UNIQUE COLDKEYS
POOL DEPTH
7.7k τ
TAO IN POOL
7D MOMENTUM
+0.0%
ALPHA vs TAO
Project activity
Dynamic pages turn crawling into a policy decision.
Desearch Web Crawl can fetch normally or render JavaScript with a bounded wait when the page needs it.
Start static. Escalate only when the content is incomplete.
#AIAgents #AISearch #Desearch https://t.co/SMBo4VXfyP
Cost attribution should not depend on response format.
Desearch exposes billing headers for JSON, arrays, text, and streams. Capture them before parsing the payload.
If a tool can spend, its orchestrator should record it.
#AIAgents #AIInfra #Desearch https://t.co/3kUVqHyZJb
Static extraction first. Browser rendering only when needed.
Desearch Crawl returns page text, with optional headless rendering for SPAs or JavaScript-populated pages.
Match the fallback to the failure mode. Do not make it the default.
#AIAgents #AIInfra #Desearch https://t.co
Engagement is not evidence.
Desearch social data can filter by minimum likes, replies, or reposts. Use those thresholds to control the sample, not to decide what is true.
Popularity is a retrieval signal, not a credibility score.
#AIAgents #AISearch #Desearch https://t.co/0Gs4
Not every search pass needs an LLM judge.
If the next step only needs candidate links, per-link scoring can be unnecessary work. Desearch lets builders skip it and keep source order for that stage.
Use intelligence where it changes the decision.
#AIAgents #AIInfra #Desearch ht
Retrieval should not turn an agent into a single-file queue.
Desearch SDKs use async patterns in Python and JavaScript. Independent searches can run concurrently, then join before synthesis.
Parallelize collection. Keep the decision ordered.
#AIAgents #AIInfra #Desearch https:
The first page is not the dataset.
If an agent summarizes before pagination finishes, a retrieval limit can become a false conclusion.
Desearch exposes pagination across web and public social data, so builders can collect before synthesis.
#AIAgents #AIInfra #Desearch https://
Mocked retrieval tests ask: does our logic work?
Live tests ask: does the world still connect?
You need both. Mocks make failures reproducible. A bounded Desearch check catches credential, network, and response-contract drift before users do.
#AIAgents #AIInfra #Desearch https
Tool calling has one important boundary.
The model proposes a function and parameters. Your application validates and executes it.
With Desearch as the retrieval tool, that separation keeps search behavior explicit, testable, and auditable.
#AIAgents #AIInfra #Desearch https:/
Agents need an error policy, not just a search tool.
Retry rate limits and transient server failures with backoff. Surface request-validation errors instead of retrying them.
Reliable retrieval starts with knowing which failures are recoverable.
#AIAgents #AIInfra #Desearch ht
More context is not always better.
A research agent needs a stopping rule: did this search add useful evidence, how fresh is it, and what did it cost?
Desearch Web Search returns freshness signals and cost metadata beside the sources.
#AIAgents #AISearch #Desearch https://t.co
Good AI search starts before the model writes.
Choose the source set, time window, and result shape first. Then generate from the evidence that survives.
Desearch lets agents scope retrieval across web, research, forums, video, and social.
#AISearch #AIAgents #Desearch https:/
SN22 now supports Bittensor 10.3.0.
The release removes a codec conflict that could break validator startup and updates the install path.
Network upgrades should be boring. That means fixing dependency edges before operators hit them.
#Bittensor #SN22 #Desearch https://t.co/q8
SN22 now supports Bittensor 10.3.0.
The release removes a codec conflict that could break validator startup and updates the install path.
Network upgrades should be boring. That means fixing dependency edges before operators hit them.
#Bittensor #SN22 #Desearch https://t.co/Qv
If docs only work after JavaScript runs, many crawlers never see them.
Desearch now serves real 200-status hubs linking 30 guide pages and 35 API-reference pages.
Docs should work for people, search engines, and agents.
https://t.co/TEAxbP88gu
#AIAgents #Desearch https://t.co
If answers only appear after JavaScript runs, many crawlers never see them.
Desearch now server-renders product FAQs and exposes crawlable hubs for 30 guide and 35 API-reference pages.
Docs should work for people, search engines, and agents.
#TechnicalSEO #AIAgents #Desearch h
Fast AI search should be measured by its tail, not just its best run.
The public Desearch benchmark now shows median and p90 response time beside source relevance, answer quality, and groundedness.
Quality and latency, on the same public scoreboard.
https://t.co/rkcOLqjYrk htt
Fast AI search should be measured by its tail, not just its best run.
The public Desearch benchmark now shows median and p90 response time beside source relevance, answer quality, and groundedness.
Quality and latency, on the same public scoreboard.
https://t.co/rkcOLqjYrk htt
Desearch AI Search now routes full-answer requests through the validator first, with Miner API fallback when that path is unavailable.
That connects product traffic to the SN22 quality loop without making reliability depend on one route.
#Bittensor #SN22 #Desearch https://t.co/
Desearch AI Search now routes full-answer requests through the validator first, with Miner API fallback when that path is unavailable.
That connects product traffic to the SN22 quality loop without making reliability depend on one route.
#Bittensor #SN22 #Desearch https://t.co/
Search becomes infrastructure when another product can build on it.
We’ve been working with @Bitcast_network on what comes next for Desearch × Bitcast.
Public details land next week.
#Bittensor #SN22 #Desearch https://t.co/0dAJNbUDMs
Long-running AI work should expose state.
Desearch forecasting returns a prediction job with reasoning, polling status, and citations.
For agent builders, “still working” is part of the product contract.
https://t.co/ukwzoq43bP
#AIAgents #Desearch https://t.co/K0lh86seyt
A source should win because it answers the question, not because it looks like the right type of page.
SN22 PR #370 uses Qwen3.5 397B and a clearer rubric. A wiki, list, or news page can score equally when it contains the answer.
#AISearch #Bittensor #SN22 https://t.co/tboO3z7O
More miners should not let one search type absorb the whole reward pool.
The new SN22 design fixes emissions at 90% for AI Search and 10% for social search. Within AI Search: 60% fast, 20% balanced, 20% deep.
Each lane competes on its own quality.
#Bittensor #SN22 #Desearch ht
Fast, balanced, and deep search are different workloads. One weak mode should not hold back the others.
SN22 PR #370 separates capacity and routing by mode, so organic traffic follows proven quality.
https://t.co/0VczKmKCqC
#Bittensor #SN22 #Desearch https://t.co/6NgubS8aFs
SELLERS DRAINED
90%
28D SELLERS EMPTIED OUT
CAPITULATION
+0.03
BOUNCE COMPOSITE (z)
HEAT 7D
0.32×
VOLUME ÷ POOL DEPTH
HOLDERS IN PROFIT
4%
TRACKED WALLETS · ENTRY BELOW PRICE
See who's buying and selling Desearch — every trade classified miner / validator / owner / outsider, plus holder cost basis and the full 23-factor screener.
OPEN THE INSIDER TAPE →