FindEpcmai Updates — What we've built,
how it works, and where it's going
We publish release notes for every meaningful update — new signal sources, pipeline coverage extensions, architecture changes, and feature launches. This is the full record.
Version 2.0 — Current Release
The Complete Pipeline
Architecture — How V2 Works
Click through the stages of the V2 intelligence pipeline below to inspect how data points are processed and what updates were introduced:
Stage 1 — Signal Ingestion
Thousands of sources are monitored continuously across regulatory portals, government databases, stock exchange filings, corporate announcements, environmental registries, and industry publications. Sources are monitored in real time — not batched weekly. Each raw data point captured is called a signal candidate — an unverified piece of information that may indicate project activity.
Source coverage extended significantly. Signal candidates are now captured across multiple geographies and regulatory frameworks simultaneously, with source credibility scoring applied at ingestion.
Version 1.0 — MVP Release
What V1 Was
V1 was the proof of concept — the answer to one question: can early project signals be captured at scale, classified reliably, and made useful to a BD team?
The answer was yes. V1 shipped a working signal ingestion and classification pipeline, a basic project database, and a manual-review feed. It was intentionally narrow.
Signal Ingestion — V1
A curated set of sources monitored on a scheduled basis — regulatory portals, select government databases, and industry publications. Ingestion ran on a fixed schedule rather than continuously.
Limitation: Scheduled ingestion introduced lag. Signals captured hours or days after publication rather than in real time.
Signal Classification — V1
Rule-based classification using keyword matching and source-type tagging. Signal type was determined by source category and keyword presence rather than learned model.
Limitation: High false positive rate on ambiguous signals. Signals from multi-topic sources were frequently misclassified.
Project Database — V1
Projects were logged as individual records linked to their originating signal. No deduplication.
Limitation: Significant duplicate rate in the database. Users saw the same project multiple times, reducing feed quality and trust.
Feed — V1
A chronological list of new project records, manually reviewed before publication. No personalisation or relevance scoring.
Limitation: Feed was not personalised. Users had to manually filter for relevant projects, reintroducing the same problem.
Alerts — V1
Weekly digest email summarising new projects added in the previous 7 days.
Limitation: Batch delivery meant users received intelligence up to 7 days after a signal was captured. No event-driven alerting.
Early project signals can be captured at scale from public sources
AI classification of signal type is viable with acceptable accuracy
Pilot users engaged consistently with the feed, validating market need
V1 MVP vs. V2 Current Release
| Capability | V1 MVP | V2 Current |
|---|---|---|
| Signal Ingestion | Scheduled, select sources | Continuous, broad source coverage |
| Classification | Rule-based, keyword matching | AI model, multi-factor scoring |
| Deduplication | None — duplicate records common | Entity resolution engine |
| Project validation | Manual review | Automated validation pipeline |
| Feed personalisation | None — same feed for all users | AI Match Feed, scored per user profile |
| Relevance scoring | Not present | 5-dimension scoring model |
| Stakeholder data | Sparse, manually curated | Automated enrichment at scale |
| Alert delivery | Weekly batch digest | Real-time, event-driven |
| Geographic coverage | Limited | Multi-geography, multi-regulatory |
What's Coming
The next release cycle will focus on expanding pipeline intelligence and introducing native CRM sync triggers:
CRM & ERP integration — pushing matched projects and stakeholder records directly into Salesforce, HubSpot, Zoho, SAP, and Odoo
Expanded signal types — hiring activity signals and corporate restructuring announcements added to the classification model
Stakeholder contact enrichment — adding verified contact details to identified stakeholder records
Team collaboration features — shared pipeline views, pursuit tagging, and team-level alert management
Want to know when new signals hit your sector?
Early access users get notified the moment new signal sources or geographies are added that match their business profile.