The element of cascade
A cascade is a fall of water that moves through stages — each level catching what the one above it passed down, each level doing work the one above could not.
That is the architecture, named. Cascadium’s core engine is a waterfall data model: a query descends through sources in priority order, cost-aware at every step. It asks the internal database first — free, instant, already verified. If the answer is not there, it falls to the next source, and the next, until the data is found, verified, and scored for confidence. Then it writes the result back into the internal database.
Which means the platform gets cheaper and faster the more it is used. Every enrichment run leaves the next one less work to do. The database is not a static asset that depreciates — it is the compounding output of the product’s own operation.
The problem Cascadium addresses
Outbound is a tool stack, not a system. A team running outbound today assembles it from parts: one tool to build the list, another to enrich it, a third to sequence the emails, a fourth to book the meeting. None of them share context — the list-builder does not know what the company sells, the enricher does not know why a lead was chosen, the sequencer does not know either. A person sits in the middle exporting CSVs, and the quality of the entire pipeline collapses to the quality of that person’s afternoon.
The deeper failure is upstream. The commodity contact databases that power most of this market are the same commodity contact databases powering every competitor — the same companies get targeted by the same people with the same stale records. Differentiation on data is impossible when everyone shops at the same store.
India compounds the gap. Global providers have thin, poorly-verified coverage of Indian companies, while India publishes an authoritative, public company registry — the Ministry of Corporate Affairs’ Company Master Data, available through Open Government Data. It is free, authoritative, and structurally unavailable to competitors built around US and EU sources, and almost nobody in this market uses it as a verification substrate.
Who Cascadium serves
- Indian SMBs running outbound for the first time
- Agencies selling services to Indian businesses
- Founder-led sales teams without an SDR function
- Revenue teams that want automation without surrendering the approval step
How Cascadium works
The flow starts with one input: your website.
Research your company. The system profiles your business from the site — what you sell, who you sell it to, how you position it. No forms, no manual ICP worksheet.
Explore competitors. Auto-detected, and used as targeting signal rather than decoration.
Define campaigns. ICP and audience segments proposed from the profile, presented as reviewable segments rather than a black box.
Find potential customers. Matching companies surfaced and verified through the identity waterfall, resolving each candidate against the MCA/OGD registry for legal existence, status, and jurisdiction before it enters a list.
Find decision makers. The contact waterfall cascades through email-finding providers in sequence until a valid or catch-all address resolves, with the provider that succeeded recorded on the record.
Write emails. Drafted per lead against the enrichment already gathered — not a template with a merge field, but copy written against what the system actually knows about that specific company.
Send emails, after you approve. The human gate. Every send is reviewed and editable before it leaves. This is not a setting; it is the design.
Book meetings. Replies are handled automatically, conversation continuing until a meeting is scheduled or the thread closes.
Learn and double down. Outcomes feed back into campaign refinement, so targeting improves against real reply data rather than assumption.
Two waterfalls, deliberately separated
Most products in this category present a single undifferentiated “waterfall” and blur two fundamentally different problems together. Cascadium separates them, because only one of them is defensible.
| Identity waterfall | Contact waterfall | |
|---|---|---|
| Question | Does this company legally exist, and what is it? | How do I reach a person here? |
| Sources | Internal DB → MCA/OGD registry → external providers | Cascading commodity email-finding providers |
| Character | Authoritative, compounding, nearly free | Commodity, per-call cost, same curve as every competitor |
| Defensibility | The moat | Table stakes |
Every record carries a confidence score and a source lineage — which layer answered, and how sure the system is. A record verified against a government registry is not presented with the same weight as a scraped guess, and the interface says so.
Multi-tenant from the architecture down. Every record carries a tenant, enforced by PostgreSQL row-level security. Shared global intelligence — companies, people, signals — is architecturally separated from private tenant data — leads, campaigns, conversations — so one customer’s outbound never leaks into another’s, while the verified company graph beneath both keeps compounding for everyone.
Human-in-the-loop is structural, not optional. Approval gates sit at list confirmation and before every send. A system that emails on your behalf without your review is not automation, it is a liability.
What makes Cascadium different
The flow above is a validated UX pattern in this category, and Cascadium does not pretend to have invented it. Copying a validated flow is a legitimate shortcut — but it means the differentiation has to live underneath it:
- Registry-grounded verification — company identity resolved against India’s own filing data, not inferred from a scraped web presence
- A compounding database — the waterfall writes back, so coverage and cost per enrichment both improve with usage
- India-first coverage where global tools are thin — not a localisation layer over a Western product, a different data spine
- Signal-based targeting over static firmographic filters
- Priced for the Indian SMB, a different price point than the market’s incumbents were built around
Protocol stage: experiment
The lab’s method runs in six rungs. Cascadium has reached experiment; everything past it is shown as not yet earned, which holds the project to the same standard as the registry’s empty slots.
- OBSERVE
- EXPERIMENTREACHED
- ANALYSEMVP-0 SCOPED
- PROTOTYPESOON
- VALIDATE
- SCALE
Release scope
MVP-0 · IDENTITY WATERFALL
Single-tenant, no billing, design partners only. One question answered: does the India data edge produce accurate, verified company data. The critical path is registry data access, confirmed before a line of application code.
MVP-1 · MULTI-TENANT PLATFORM
Tenant isolation via PostgreSQL RLS, lead scoring, pipeline view, usage metering against the credit model — built once MVP-0 has a paying yes.
V2 · THE ACTION LAYER
AI SDR, conversation AI, WhatsApp, calendar booking, RAG knowledge base, automated ICP optimisation, expanded provider set, public API.
V3 · PROPRIETARY INTELLIGENCE
Owned company signals, predictive buying intent, enterprise SSO, data marketplace and API, voice AI.
Go to market
First segment: Indian SMBs and agencies. The MCA/OGD edge lowers customer acquisition cost in exactly this segment, because the pitch is verifiable in the first demo — the system produces registry-accurate results on companies the prospect already knows.
Pricing is hybrid: a subscription tier (Starter, Growth, Business, Enterprise) plus usage credits metered across search, enrichment, generation, research and communication, so light users are not subsidising heavy ones and the compounding-database economics pass through to price.
Build sequencing follows the same discipline: prototype and validate before committing full build time. MVP-0 exists to answer one question — does the identity waterfall work — and nothing else. Contact enrichment, scoring, CRM and the full action layer are deliberately deferred until that answer is yes.
Stack and infrastructure
- MCA Company Master Data via India’s Open Government Data platform, forming the identity data spine
- Next.js and React on the front
- A modular monolith with worker services rather than premature microservices
- PostgreSQL with row-level security for tenant isolation
- pgvector for semantic matching
- Redis and BullMQ for the job queues
- OpenSearch when scale demands it
- S3-compatible object storage
Positioning: two waterfalls, not one
Company identity resolves through India’s MCA and Open Government Data registries — authoritative, compounding, and structurally unavailable to global competitors. Contact-finding cascades through commodity providers like everyone else. Only the first one is a moat, and Cascadium is built so the moat feeds the commodity layer rather than the reverse.