One workspace. Every step of content creation.

Deaku replaces the tangle of Notion, Google Docs, Buffer, and separate analytics tabs. It is a single environment that takes a creator from raw idea to published piece to performance data — without switching context once.

How Deaku works

Every step connects to the next. No dead ends, no exports, no reformatting.

01

Capture layer

An always-on idea inbox accepts text, voice-to-text (via Web Speech API), and URL clips. Every item is timestamped and auto-tagged by inferred topic using a lightweight classification model running client-side.

02

AI brief engine

Deaku calls our hosted inference API (sub-100ms p95 latency) to generate a structured brief: working title, 3-sentence hook, full outline with H2s, and a suggested format (article, short-form video, newsletter). The model is fine-tuned on 14,000 creator briefs, not general web data.

03

Distraction-free editor

A Prosemirror-based editor with no persistent sidebar. Markdown shortcuts supported. AI co-writing is opt-in per session — press Cmd + J to summon a continuation, or leave it alone and it stays silent.

04

Multi-channel publish queue

Connect via OAuth 2.0 to YouTube (Data API v3), Instagram Graph API, LinkedIn Pages API, X API v2, TikTok Display API, and Ghost Admin API. Each platform has its own character limit validator, image dimension checker, and scheduled delivery queue backed by a Redis job runner.

05

Unified analytics

Platform-specific metrics (views, watch time, engagement rate, follower delta, link clicks) are normalised into a single data model and stored in a time-series Postgres schema. Dashboards refresh every 4 hours. Revenue tracking available for YouTube Partner and Ghost paid newsletters.

What the platform does, precisely

We describe every capability in technical terms because creators deserve to know exactly what they are using.

Voice-to-Idea Capture

Web Speech API captures voice notes directly in the browser — no app download required. Audio is transcribed locally (no server round-trip), cleaned with a punctuation model, and added to the idea inbox. Works on Chrome and Edge on desktop.

Use case: recording an idea while walking or between meetings without reaching for a notes app.

Tone Analysis

Before generating a brief, Deaku analyses the last 20 published pieces from your connected accounts. It extracts sentence length distribution, vocabulary frequency, and reading level (Flesch-Kincaid grade). The brief engine uses these as soft constraints so generated content sounds like you.

Result: briefs that don't require heavy rewriting because they start from your baseline.

Optimal Timing Engine

Deaku tracks when your past 90 days of posts received the most engagement by platform and audience timezone. The scheduler surfaces the top 3 posting slots per platform per week. You override with one click — it's a suggestion, not a lock.

Built on engagement-per-impression, not vanity metrics like total likes.

Content Repurposing Queue

Mark a long-form piece as a repurposing source. Deaku automatically generates a short-form version (under 300 words), a thread structure for X, and a newsletter section. Each output is editable before it enters the publish queue.

Reduces repurposing time from 40 minutes to under 8 minutes per piece, based on beta user logs.

Platform Integrations

Every integration uses official OAuth 2.0 flows — Deaku never stores passwords. Webhook support allows real-time sync of published status and comment alerts back into the workspace.

YouTube Instagram LinkedIn X / Twitter TikTok Ghost Zapier Notion (import)

Security and Privacy

All workspace data is encrypted at rest with AES-256. Credentials (OAuth tokens) are stored in a separate secrets vault, isolated from application database. Content drafts are never used to train any model — confirmed by DPA agreement with our inference provider.

Data residency: EU (Ireland). GDPR compliant. SOC 2 Type II audit scheduled for Q3 2025.

Built to handle daily publishing at volume

API Performance

  • Brief generation: <100ms p95 (edge inference)
  • Scheduler API: <200ms p99
  • Analytics refresh: every 4 hours per connected account
  • Uptime SLA: 99.5% monthly

Data Architecture

  • PostgreSQL with time-series extensions for analytics
  • Redis job queue for scheduled publishing
  • S3-compatible object storage for media assets
  • AES-256 encryption at rest, TLS 1.3 in transit

Supported Platforms

  • YouTube Data API v3
  • Instagram Graph API
  • LinkedIn Pages API
  • X API v2 (OAuth 2.0 PKCE)
  • TikTok Display API & Ghost Admin API v5

See it in a 20-minute live demo

We walk through a real content brief to published article — no scripted slides, just the actual product.

Book a Demo