Without Context Filter
Weavi Admin Portal
Email & password
Sign in with your registered email and password.
One-time email link
No password required. We'll send a secure sign-in link to your authorised email address.
Signed in as
My workspace
Your account, organisations and API access.
My usage & measured benefit · last 30 days
—API requests
—Input tokens
—Tokens not sent onward
—Measured reduction
Usage appears after customer-key requests are aggregated.
Platform overview
—Organisations
—Active API keys
—Pending approvals
—Requests (30d)
⏳ Pending portal access approvals
These users have signed up but not yet been granted portal access.
Top organisations · last 30 days
Service health
API server
Checking…
Redis
Checking…
Key verifier
Checking…
Quick links
My profile
Update your display name, avatar and password.
Change password
Awaiting portal access approval. Your account is authenticated. No customer data or API keys are available until a super-user approves your access.
Current organisation
Create your first organisation
You don't have an organisation yet. Create one to manage trial access and API keys.
No organisation yet. Set up an organisation on the Organisation tab before creating API keys.
⚠ Save this API key now
For security, the complete key is shown once and is not stored recoverably.
Create a customer API key
Customer keys (wv_) allow the v3f, v3b and v3e algorithms with controlled trial limits. No access to internal test endpoints.
Create an internal API key Super-user
Internal keys (wvi_) have access to all algorithms including diagnostic and test endpoints. For internal use only — never share with customers.
API keys
Customer API keys
Customer organisations grouped with their primary contact and API keys.
Internal role management
Super-admin only. Assign internal admin or super-admin access independently from customer organisation roles.
All API keys — platform-wide
All customer and internal keys across every organisation. Enable, disable, or edit limits on any key.
Platform usage
Period:
—Total requests
—Tokens saved
—Avg reduction
—Avg latency
Requests per day
Tokens saved per day
Usage by organisation
Organisation rate limits
Set the overall API rate limits for any organisation. These are the maximum limits customers can allocate to their individual keys.
IP allowlist slot limits
Set the maximum number of IP entries (hosts or CIDR ranges) a customer can add across all their keys. Tiers: Trial=3, Starter=10, Professional=50, Enterprise=250, Unlimited=0.
IP / CIDR blocklist
Block overrides whitelist. Any IP matching a blocklist entry is denied access regardless of per-key IP whitelist settings. Use CIDR notation for ranges (e.g.
198.51.100.0/24).
Select a customer
Choose a customer from the left, or add a new one.
Customer
0Contacts
0Open reminders
NeverLast contacted
—Est. monthly tokens
Contacts
Activity
Open reminders
Email history
Add customer
Create the business and its primary contact together.
Add contact
Add a contact to the selected CRM record.
Add activity
Record an interaction or note against this CRM record.
Add reminder
Schedule a follow-up for this CRM record.
Compose email
Send an email to a contact on this CRM record.
Email Gateway
Transactional delivery for magic-link invitations and password recovery. Credentials are held only in Supabase Auth configuration and are never sent to this browser.
—Provider
Checking…Configuration
—Sender
Send recovery test
Sends a real single-use password-recovery email through the configured gateway.
✚
SignalLens Safety IntelligencePowered by Weavi Context Filter · OpenFDA
Adverse-event signal exploration
Find the safety signal.
Lose the context noise.
Explore reported drug-event patterns while comparing a full source corpus with an optimized, evidence-focused context.
Aspirin · elderly bleeding1M data tokens4,000-token contextGemini 2.5 Flash
Build a live OpenFDA comparison
What reported factors are associated with bleeding, and which patient groups appear most affected?
FDA analysis in progressPreparing the configured test…
0sNo analysis has been run
Review the current configuration above, adjust it under Settings if needed, then run the live safety analysis.
Measured benefit
—Records removed
—Prompt tokens saved
—Estimated cost saved
—Selection time
Importance-tier retention
Answer comparison
With Context Filter
Represented-scale test
This test uses actual FDA records with proportionally weighted token counts. It measures algorithm behaviour without claiming actual ingestion or LLM processing of the represented volume.
Selected record samples
Demo and educational use only. OpenFDA adverse-event reports do not establish that a product caused an event. A single run does not prove answer-quality equivalence.
Real SEC evidence · paired methodology
Scientific Evidence Lab
Run repeatable known-item retrieval experiments against real SEC 10-K records. Compare the full-corpus model baseline, pure selector recall, and the selector-plus-model pipeline without substituting simulated outcomes.
1 · Test mode
2 · Source data
3 · Test design
4 · Review
Evidence run in progressFetching real SEC filings and preparing paired targets…
0s∑
No evidence run yet
Configure a reproducible test above. Results will show observed retrieval outcomes, confidence intervals, corrected significance tests, ranking quality, token use and raw evidence.
Saved run history
Completed and failed runs persisted by the protected control plane.
Open this tab or refresh to load saved runs.
Observed result
—Selector recall
—Selection MAP
—Selection MRR
—Token reduction
—Test duration
Three-stage comparison
Retrieval metrics
Statistical evidence
Observed paired outcomes, uncertainty and multiplicity controls—not decorative test counts.
Information and power diagnostics
Sampled targets
| Tier | Position | Importance | Baseline | Selected | Pipeline | Preview |
|---|
Raw reproducibility evidence
Complete server response for audit and independent review.
◇
DESC document intelligence demo
A second domain-specific Context Filter showcase is being prepared.
Coming soonSaved comparison models
Save five OpenRouter model IDs. These become the graphical model choices in every FDA analysis wizard.
FDA demo defaults
Review the current test setup. Use Configure analysis to change any value through the guided workflow.
Research sceneAspirin · elderly · bleeding
ModelGoogle Gemini 2.5 Flash
FDA data budget1M token equivalent
Context budget4,000 tokens
SelectionAutomatic routing · automatic chunking
QuestionWhat reported factors are associated with bleeding?
Configuration is held only in this browser session. Large data and completion budgets may increase provider cost and response time.