000 Content intelligence platform

Read the world like one room.

InsightStorm turns the open web — news, filings, Telegram channels, PDFs, structured feeds — into one queryable, attributed corpus. Configure your sources and taxonomy. We run the rest.

Sources
0
Docs / day
0
Languages
0
Tenants live
0
STORM · LIVE

001 Gap

Every team is drowning in signal,
starving for meaning.

The open web is where the answers live — and where the big platforms cannot, or will not, reach. Three doors, three problems.

  • Blocked by policy

    Gemini · Copilot · OpenAI Enterprise

    Cannot crawl websites their customers don't own. Domain verification is a hard policy gate that protects Search — they will not break it. Open-web ingestion is structurally off-limits.

  • Expensive by design

    Recorded Future · Bloomberg · Palantir Foundry

    Closed feeds priced from $50k floors. Foundry-class deployments land at $1M+ with a Forward Deployed Engineer team. New sources mean change orders, not config.

  • What InsightStorm is

    Headless SaaS, configured by you

    Open-web, third-party, multilingual sources — declared per tenant, ingested in one pipeline, normalized into five forms. Hours to first data. No FDE team in the loop.

002 Capabilities

One pipeline,
three movements.

Collect everywhere. Read structure. Answer with proof. Six capabilities, one platform — no Forward Deployed Engineers required.

  • Listen everywhere

    Web articles, RSS, JSON APIs, encrypted messaging channels, PDFs with right-to-left OCR, even legacy form-based web catalogs that no SaaS crawler reaches. Declarative config per tenant; no platform release required.

    • RSS
    • Web
    • PDF + OCR
    • Telegram
    • JSON APIs
    • ASP.NET forms
  • Autonomous discovery

    A Gemini-driven agent runs hourly per tenant — hunts new candidates on the open web, validates relevance against your taxonomy, queues the keepers for one-click approval.

  • Extraction in any language

    Right-to-left and complex non-Latin scripts first-class. PDFs go binary-to-JSON in a single model call. Original text preserved; English translation, summary, entities indexed for cross-lingual search.

  • Five forms of meaning

    Every sentence in every source resolves into articles, events, signals, positions, or records. One analytical layer across all five — search, aggregate, group-by on any field path. From a GPS-frame telemetry packet to an M&A headline, one model.

    • article
    • event
    • signal
    • position
    • record
  • Semantic + analytical search

    BM25 keyword and pgvector similarity in a single hybrid query. Filter, aggregate, group-by on any data.* field path. One engine, every form.

  • Answer with proof

    Natural-language questions over the whole archive. The analyst chatbot knows your taxonomy, your source counts, your data.* paths. Every claim cites a URL.

003 Source discovery

It hunts.
You approve.

An autonomous Gemini agent runs hourly per tenant — looking for sources you don't know about yet. Scores them against your taxonomy. Surfaces the keepers for one-click approval. No competitor has this productized.

discovery@storm:~$ hunt --tenant=$TENANT --taxonomy=$CODEBOOK RUNNING
candidates
    scoring
      approved · queued
        last hunt · 00:00:00 0 candidates inspected this hour

        004 Process

        How a tenant
        goes live.

        Four steps from blank slate to running storm. Days, not quarters.

        1. 01

          Declare your taxonomy

          Codebooks, extraction profiles, dashboards, chatbot prompts — declarative JSON per tenant. No engineer required.

        2. 02

          The storm finds sources

          Source discovery runs hourly, proposes new feeds. You approve the keepers in the admin UI.

        3. 03

          Pipeline ingests and extracts

          Gemini produces title, summary, entities, classifications. Multilingual in, English index out, full provenance preserved.

        4. 04

          Search, chat, alert, decide

          Analyst surface goes live: hybrid search, BI dashboards, maps, AI chat, alerts — all scoped to your tenant.

        005 Golden Record & Entity Resolution

        Many claims.
        One Golden Record.

        Unstructured open web intelligence is noisy and disjointed. InsightStorm automatically resolves conflicting, multi-source statements into a single, high-confidence Golden Record using declarative survivorship rules.

        unresolved claims (input) 3 sources
        the-hacker-news 18:03:28

        "Threat group Nexus-921 is an affiliation: Nexus-921 group targeting DeFi."

        Confidence: 0.90 Method: llm_entity_fact
        cisa-advisory 18:04:41

        "The threat actor is a role: state-sponsored hacking group operating from Region-Alpha."

        Confidence: 0.95 Method: rule_engine
        ngo-watchdog 18:11:28

        "Region-Alpha actors deploy malware in Europe from location: Region-Alpha."

        Confidence: 1.00 Method: trusted_analyst
        unification engine (process)
        IDLE
        G

        Nexus-921

        NORP / THREAT_ACTOR
        golden record
        affiliation
        Nexus-921 the-hacker-news (90%)
        role
        state-sponsored hacking group cisa-advisory (95%)
        location
        Region-Alpha ngo-watchdog (100%)
        resolved graph (output) graph active
        entity statement resolution graph
        Hover over nodes to inspect statement lineage
        SHOWCASE LOOP: highest_confidence

        006 Atlas

        If it's published,
        we hear it.

        Numbers reflect live production tenants today — across multiple analyst-driven verticals from investigative intelligence to market-signal research. The atlas grows hourly.

        0 active sources, growing hourly
        • RSS 0 feeds
        • WEB 0 sources
        • PDF 0 docs / day
        • TELEGRAM 0 channels
        • JSON·API 0 endpoints
        • FORM 0 catalogs
        • Project Nimbus
        • GCP
        • SOC 2
        • SSO
        • DPoP enforced
        • Per-tenant isolation

        006 Compare

        Where everyone else stops,
        InsightStorm starts.

        Three competitors, one platform-of-record. What ships, what costs, what takes.

        Gemini / Copilot Palantir Foundry RF / Babel InsightStorm
        Open-web ingestion Blocked by policyCustom code per sourceCurated only Yes
        Customer-defined sources NoYes — with FDE teamNo Yes — self-serve
        Autonomous source discovery NoNoManual analyst work Yes — hourly per tenant
        Per-tenant taxonomy NoYes — with FDE teamNo Yes — declarative
        Multilingual + RTL LimitedYesLimited First-class
        Pricing $30–100 / seat$1M+ contracts$50k+ floors Contract-based SaaS
        Time to first data Days — internal docs4–8 weeksOut of your control Hours

        008 Workspaces

        One pipeline.
        Four workspaces.

        InsightStorm is domain-agnostic. By changing a tenant's declarative codebook and prompt templates, the entire dashboard re-configures itself for any investigative vertical. No coding required.

        Cyber & Threat Intelligence

        tenant: security_desk_prod
        codebook.schema.json JSON
        // Configuration loads dynamically
        resolved_stream_feed POLLING ACTIVE

        Production Scenarios & Operational Impact

        009 Pilot

        Read the world
        like one room.

        30-day pilot. One desk, one watchlist, one decision per week — measured against your current process. Sign in to access your tenant, or talk to us about provisioning one.