LensAIFrom instrument to network.
The build order is deliberate: get the live pipeline and the non-advisory read right first, decouple cost from users, then widen coverage and open the layer up. This is where we are and where we're going.
Pipeline first
Every phase strengthens the live-data pipeline before it adds surface area. The pipeline is the product; features are downstream of it.
Cost decoupled from users
The expensive work, covering popular assets, is pre-computed and shared, so serving one more user costs almost nothing.
Non-advisory, always
No phase ever ships a buy button. Everything we build assesses signals and presents both cases; the decision stays with the user.
The instrument
Prove the core loop: sign in without friction, gather live data, and return an attributed read.
- Wallet sign-in (SIWE): a signature, no gas, no PII
- Live market pipeline (primary source + fallback)
- Six-section, non-advisory analysis template
- Streaming reads with the POSITIVE / MIXED / NEGATIVE call
- Two free analyses per wallet, enforced server-side
The desk
Turn the instrument into a place you return to: fast, cheap, and with memory of what you've researched.
- Per-ticker cache + scheduled top-token pre-compute
- Follow-ups resolved from stored fields, no new model call
- Pre-fetched news for popular assets to kill the search fee
- Research history, watchlist and reopenable sessions
- One-action account and per-session deletion
The research desk & its agents
Read the whole market, not one token at a time, then make that reading something you can use.
- Built: immutable point-in-time capture across funding, basis, macro, flows, depth and news
- Built: every stream normalized against its own history, regime-tagged, joined on one clock
- Built: curated mechanism graph and a pure-arithmetic divergence engine
- Built: fail-closed narration: every number resolves to a store row or the claim is dropped
- Built: the agent API: ask, what changed, standing watches, claim check, the record, tool endpoint
- Built: the desk interface for the agents, and the always-on deployment
Breadth & depth
Cover more of the market, and go deeper on the assets that deserve it.
- Long-tail token coverage via strictly-capped live search
- Model tiering: escalate to stronger models for complex assets
- Richer on-chain structure: holder maps, unlock calendars
- Portfolio-aware context, computed live and never stored
- Shareable, timestamped read snapshots
The network
Open the desk up to integrations, and to the community that uses it.
- Typed claim extraction from the corpus: who said it, with what incentive
- Outcome scoring of divergence flags in the public calibration record
- Community coverage requests and prioritization
- Token utility and governance (see tokenomics)
The standard
Make LensAI the reference layer other products build on top of.
- Embeddable read widgets and partner integrations
- Multi-chain and cross-asset comparison
- Historical read archive to study how signals aged
- Transparency dashboard for methodology and sources
Timelines describe intent, not commitments. Priorities shift as the pipeline and the market teach us where the value is.