Public overview — source code is private. This repo describes what the product does and how it's built.
Executive search firms spend hours turning raw candidate material — CVs, Hogan/psychometric reports, interview notes — into beautifully formatted, on-brand client deliverables (assessment reports, executive profiles, shortlists). It's high-skill, repetitive, error-prone work.
Formatix turns that raw material into client-ready, brand-perfect documents in minutes, while keeping the consultant's judgment at the centre:
- Notes-driven reports — consultants own the assessment; Formatix formats it faithfully and fits each section to a consistent length, with no AI "drift" or invented content.
- Multi-format generation — pixel-accurate PPTX, DOCX and XLSX output that matches each firm's exact template, fonts and branding.
- Document intelligence — OCR of scanned inputs, face-aware profile-photo extraction, psychometric chart rendering, and automatic section selection.
- Per-run tailoring — role, report voice (1st/3rd person), and key criteria configurable per candidate.
- MCP server — an
@formatix-ai/mcpintegration lets Claude, ChatGPT, Perplexity and Cursor drive the platform directly.
Status: Live in production with paying customers.
Uploads (CV · Hogan · notes) ─► Extraction (multi-LLM, schema-validated JSON)
│
multi-model routing + fallback (Claude · Gemini · GPT)
│
Template fill engine (PPTX/DOCX/XLSX) ─► brand post-processing ─► client-ready file
│
MCP server · REST API · web app
Python · Django · Django REST Framework · MySQL · multi-LLM (Anthropic Claude · Google Gemini · OpenAI GPT) with routing & fallback · python-pptx / python-docx / openpyxl · OpenCV (OCR & face detection) · Pillow · 2FA & OAuth · Azure · Apache · GitHub Actions CI/CD · Model Context Protocol server.
Built by Dominic Gonsalves · formatix.ai · open to contract / fractional AI roles