Posto AI: Facebook posts rewritten in your own voice

01 / Overview

About Posto AI

02 / The challenge

What Posto AI needed to solve

Generic AI rewrites all sound alike, and Facebook audiences notice. Posto had to be personal, careful and affordable to run at the same time.

1. 01

   Rewrites had to match one person's formality, emoji habits, sentence length and vocabulary instead of a stock AI voice.

2. 02

   The tool had to handle five different kinds of text, each with its own norms on Facebook.

3. 03

   Hateful or misleading drafts had to be refused rather than polished into something more convincing.

4. 04

   Selling dressed up as casual chat needed to be detected, and sensitive topics like grief and politics needed a gentler touch.

5. 05

   A free level had to exist next to paid tiers without letting model costs or abuse run out of control.

03 / What we built

How we built Posto AI

We built a web app where every request runs through a fixed nine-step pipeline, with checks before and after the model writes anything.

• 01

  Nine-step rewrite pipeline

  Each request passes rate limits, a repeat check, a policy scan, a promotional score and a sensitivity check before the rewrite, then a second scan of the output. Nothing reaches the user without being checked twice.

• 02

  Style learning from your posts

  Users add two to five of their own posts, and Posto measures formality, emoji use, sentence length and vocabulary. Rewrites then sound like the user, not like a chatbot.

• 03

  Rewrite studio

  Users pick a content type, choose Standard or Promotional mode, and adjust tone sliders for casual to formal, conservative to punchy, brief to detailed and serious to warm. The draft and the rewrite sit side by side with a separate hook option.

• 04

  Policy and sensitivity checks

  Fixed patterns refuse six kinds of content before any model call is made. Drafts about grief or politics are handled more gently instead of being punched up.

• 05

  Promotional intent scoring

  Drafts that score above 65 out of 100 for promotional intent stop with a prompt to switch to Promotional mode. Hidden hard selling gets flagged instead of slipping through as chat.

• 06

  Engagement score for Pro

  Pro users get an engagement score for each rewrite along with a suggestion to improve it. This helps page owners decide what to post before they publish.

• 07

  Tiered plans with Stripe

  Visitors can try five rewrites a day with no account, free accounts get twenty, and Plus, Pro and Teams plans are billed through Stripe checkout, a billing portal and a signed webhook.

• 08

  Marketing site and dashboard

  A landing page, pricing page, sign in and a dashboard for new rewrites, rewrite history and style profile, all working on desktop and mobile.

04 / The results

What changed for Posto AI

The finished product covers the full rewrite flow, the safety layers, style learning, tiered billing and a marketing site.

• Every rewrite is checked before and after the model runs, so the tool polishes what people mean to say without handing back text that breaks the rules.
• Users get rewrites shaped by their own past posts instead of a generic AI tone.
• Fixed pattern checks refuse clearly bad content before a model call, which keeps costs down on requests that should never run.
• Rate limits by account type let Posto offer a real free level while paid plans cover heavier use.