Folocard
First-Touch Mini CRM

From the first handshake to a reviewed next action.

Folocard is not a business-card scanner. It is a proactive local assistant for field professionals, founders, and operators. It collects field evidence, builds a sourced person dossier, and prepares reviewed follow-up actions on hardware you control.

Get Folocard on Google Play Explore Features

Folocard executes on-device AI inference and stores contacts locally without remote cloud data brokers. Follow-up drafts are prepared for your review and approval.

Operating Architecture

Core Operating Principles

Three foundational design commitments that govern how Folocard handles data and compute.

Local Compute Hierarchy

We organize compute strictly: Deterministic > Probabilistic > Human. Fast, zero-cost rules and optical recognition resolve regular fields first. Local language models resolve noise and ambiguity. Human attention remains reserved for outcome, voice, and intent.

No Canned Templates

Folocard does not insert generic email template fallbacks and pretend they are intelligent drafts. When a local model is unavailable, drafting stays pending while your contact and context remain safely recorded.

Clear Network Disclosures

Inference and dossier storage stay on-device. When public-web research is enabled, search queries and target URLs reach those providers. Model downloads connect directly to Hugging Face, and Google ML Kit transmits SDK diagnostics.

The Five Core Stages

How Folocard Works in the Field

A continuous workflow built additively on top of private device primitives.

01
Evidence Collection

Capture

Capture first-touch evidence at the speed of the interaction. Use the device camera for physical business cards (single card or fast batch mode), import digital vCards, record meeting voice memos, or type brief field notes.

  • On-device camera capture with real-time frame sharpness analysis.
  • Google ML Kit OCR processes text locally on your device.
  • Batch mode keeps the camera active to scan stacks with session cards-per-minute pacing.
  • Support for manual notes, meeting audio transcripts, and imported documents.
02
Relational Grounding

Context

A business card by itself is incomplete. The value lies in the person-to-person interaction: where you met, what you discussed, what mutual commitments were made, and which event calendar context applies.

  • Bounded calendar event lookups with explicit user-granted time windows.
  • Interaction tags for conferences, investor meetings, customer visits, and cafes.
  • Preserves original user observations without overwriting evidence.
03
Iterative Investigation

Sourced Dossier

Folocard follows an open-source intelligence (OSINT) approach. It builds a cumulative, sourced dossier on the person and organization using local AI reasoning and private device storage.

  • Optional public-web research sends targeted queries and fetches public URLs.
  • Local AI extracts career milestones, company background, and mutual interests.
  • Distinguishes verified facts from unconfirmed claims and recorded unknowns.
  • Reasoning and the resulting Markdown dossier stay in private local storage.
04
Human-in-the-Loop

Reviewed Draft

Folocard prepares a personalized first follow-up message synthesized directly from the contact's dossier and your interaction notes. No background automation sends messages without your review.

  • Drafts run on local language models on your device hardware.
  • Thumb-first review UI: accept, tweak, or reject prepared suggestions quickly.
  • Zero silent sending: opens your chosen email or messaging client for review.
05
CRM State

Next Action

Turn each first touch into a durable first-touch CRM state. Folocard tracks the contact, organization, thread history, sequence status, and next scheduled commitment.

  • Structured objects: Contact, Organization, FirstTouchThread, NextAction.
  • User-directed handoff to device calendar and system contacts.
  • CSV and vCard export for spreadsheets, CRM import, and address books.
  • Maintains relationship momentum without enterprise CRM complexity.
Technical Integrity

Honest Privacy Boundaries

We disclose real system behavior rather than making unsupported marketing claims.

Local AI Processing

AI inference and dossier storage operate directly on device hardware. We do not operate remote inference clusters, and third-party data brokers remain disabled in Folocard 0.11.

Public-Web Research Queries

When optional web research is enabled, search queries and requested web page URLs leave the device to reach providers (such as Brave Search) and destination websites.

Model Downloads

When downloading an on-device AI model, your device connects directly to Hugging Face. Hugging Face receives standard IP network connection data and the requested file name.

Google ML Kit Diagnostics

Google ML Kit performs optical character recognition locally on your hardware. Separately, the Google ML Kit SDK transmits technical performance and diagnostic metrics to Google over HTTPS.

Storage & Sandbox

Data resides in private operating system storage. We do not claim every file is individually encrypted with custom layers; the platform sandbox protects local files.

Human Review by Design

Folocard prepares actions and follow-up drafts for your inspection. It never sends messages silently in the background; you review, adjust, and send each draft through your own messaging client.

Ready for a principled first-touch assistant?

Capture first-touch contacts, build sourced dossiers, and draft reviewed follow-ups on Android.

Get Folocard on Google Play View All Features Read Architecture

Questions or inquiries: feedback@folocard.com. Follow-up drafts open in your preferred email client so you review and approve every message.