Contact

The Sales Stack for Scalable Processes

Small companies can't afford a full sales department. But without data-driven processes, there's no way to stay on top of things. We build the software that makes it possible.

Five building blocks, one system

Each tool solves a specific problem. Together, they form an end-to-end process from research to outreach to customer care.

Data enrichment

Company Tool

Imports companies and contacts, enriches them automatically from 8+ sources, and syncs with your CRM. No more manual research.

Details & Tech-Stack

Data enrichment is the starting point of every sales process. The Company Tool imports lists of companies and contacts — via CSV, URL, or AI agent — and enriches them automatically.

  • Multi-source enrichment from 8+ sources (websites, LinkedIn, Xing, Wikidata, Hunter.io, email validation)
  • Deep dive module: extracts structured data from websites (technologies, team, products, contact details)
  • AI agent for autonomous company research on the web
  • Flexible views with filtering, sorting, and search across enriched data
  • TTL-based freshness control — sources are only re-queried when needed
PythonFastAPIPostgreSQLRedisCelerySolidJSLangGraphAnthropic Claude
Document intelligence

DocChat

Stores internal documents in a vector/graph database and makes your entire knowledge base searchable through AI chat or API.

Details & Tech-Stack

DocChat turns internal documents into searchable knowledge. Files are automatically chunked, embedded, and stored in a graph with similarity relationships.

  • Automatic indexing of Markdown, PDF, DOCX, Excel, YAML, JSON, and images
  • OCR for scanned PDFs — even non-digital documents become searchable
  • Image analysis: OCR text recognition and LLM-generated image descriptions
  • Real-time monitoring of local folders or Seafile cloud storage
  • Hierarchical chunking with heading awareness and context breadcrumbs
  • Semantic search with path filtering and time filters
PythonNeo4jMistral AITesseract OCRWatchdogSeafile API
Email intelligence

MailChat

Indexes email accounts (IMAP), extracts contacts and conversation threads, and makes all email knowledge fully searchable, attachments included.

Details & Tech-Stack

MailChat indexes email accounts via IMAP and makes conversations, contacts, and attachments searchable. Every email is summarized and embedded into a knowledge graph.

  • Multi-account IMAP integration with incremental sync
  • Automatic contact and thread extraction
  • All attachments processed through DocChat's advanced ingestion pipeline (incl. OCR and image analysis)
  • Email management via API (flags, move, delete, drafts)
  • Conversation disentangling (quoted replies are detected and separated)
PythonNeo4jMistral AIIMAPFastAPI
LinkedIn → CRM

LinkBridge

Imports LinkedIn connections directly into your CRM with deduplication, label management, and a review workflow.

Details & Tech-Stack

LinkBridge bridges the gap between LinkedIn and CRM. Connections are exported as CSV, uploaded, automatically matched against existing CRM contacts, and imported.

  • Built for well-organized import of 10,000+ contacts
  • 3-tier matching: LinkedIn URL → email → name
  • Batch processing with review workflow and label management
  • Multi-step import — import in rounds without re-uploading
  • Automatic organization mapping with name normalization
PythonFastAPIPostgreSQLSolidJSPipedrive API v2
AI orchestration

API Browser

Connects all tools via the Model Context Protocol (MCP). A single AI agent accesses CRM, documents, emails, and data enrichment at the same time, within one conversation.

Details & Tech-Stack

The API Browser is the orchestration layer. It implements the Model Context Protocol (MCP) and makes every connected software platform controllable by AI agents.

  • MCP server with 4 tools — no matter how many services are connected, no tool explosion
  • Agents communicate simultaneously with all connected platforms — full read and write access to the complete API
  • Exact and semantic search (where supported by the respective API)
  • Scripting: agents can write scripts with full API access across all platforms
  • Fine-grained permission system for precise access control per route and service
  • Authorization and API keys managed centrally — via agent or web interface
PythonFastAPIMCP SDKSQLiteNeo4jOAuth 2.0YAML manifests

How it works

The architecture follows a clear principle: one AI agent as the central interface, the API Browser as mediator, and specialized services for each task.

Browser Web UI Console CLI AI Chat Agent MCP Server Python Server API Browser Configuration & Auth Permissions Read & Write Pipedrive REST API Company Tool REST API DocChat Knowledge DB MailChat Knowledge DB ZeroBounce REST API Cloud Storage REST API Email IMAP Calendar CalDAV

The API Browser can be accessed in four ways:

AI Chat Agent
The primary access point for users. The AI agent can either explore data directly or write Python scripts for more complex tasks.
Browser
Primarily for users to manage API keys and authorizations, enable/disable services, and configure permissions.
Console
For admins to manage deployments and system setup via scripts.

The architecture is infinitely extensible with additional API integrations. Not shown but already implemented are Mistral and OpenAI. Planned integrations include Plane (project management), SevDesk (accounting), Rocket Chat, Apify, and Google (Gmail, Drive, Docs, Sheets).

What the technology enables

The stack is more than the sum of its parts. The combination of an integrated data foundation and AI orchestration opens possibilities that aren't achievable with individual tools.

End-to-end software stack for the entire sales process

The entire sales process in one system: market research, data enrichment, outreach, and customer care, without media breaks.

AI-powered flexibility

Instead of rigid workflows, the stack adapts. AI agents navigate the software platforms contextually, responding intelligently to the task at hand.

AI scripting: the best of both worlds

Tools like n8n are strong at integration but weak at data processing. AI code tools like Claude Code are strong at logic but struggle with authentication and infrastructure. The API Browser combines both and enables complex, cross-platform workflows through natural language.

Example: deep dive workflow

"Trigger a Company Tool deep dive for every company in the CRM. Create all found persons that have an email address. Give them the label 'Deep Dive'." Flows like these used to take developers days or weeks to build. Now users can create them in minutes.

Interested?

Contact us for a personal walkthrough of our AI Sales Stack — we'll show you how it works in your setup.

Get in Touch

Or contact us directly:

info@frankeundschwarz.de