Peak

• 

• 

• 

• 

Inspiration

Creators already have analytics, but analytics mostly show what happened. They do not clearly explain what to repeat, what to change, or what to write next.

When a post performs badly, the creator is often left guessing. I built Peak to connect the past week of content with the next week of ideas.

What Peak does

Peak turns a creator’s post history into a practical content plan.

A creator uploads a real CSV export from X analytics. Peak validates the file, calculates engagement rate from impressions and engagements, and compares each post against that creator’s own baseline.

It separates stronger posts from weaker posts and shows likely signals behind weaker performance, such as:

• A weak opening
• Excessive length
• Missing hashtags
• A weaker posting time
• Performance below the creator’s own baseline

Peak then studies the stronger posts and extracts a repeatable recipe:

• Opening patterns
• Ideal post length
• Hashtag count
• Stronger days
• Stronger hours
• Media usage
• Examples of the creator’s writing style

The creator can send that recipe and the winning examples to Featherless. Peak returns editable drafts for the next week, with a reason and suggested time for each draft.

The creator can edit the drafts directly in Peak, copy one draft, copy the complete set, or download the results as TXT or CSV.

How I built it

Peak is built with Next.js, React, TypeScript, and Tailwind CSS.

The CSV parser and analysis engine are deterministic TypeScript modules. Peak recomputes engagement rate, finds the creator’s median baseline, classifies posts by quartile, and extracts patterns from stronger posts.

Featherless is used for the final drafting step. Peak sends the extracted recipe and winning examples to the Featherless API, validates the returned response, and displays the drafts in the product.

The application is deployed on Vercel and is available as a live product.

Challenges

The main challenge was building a useful automation workflow without depending on a paid social media API.

Instead of creating a fake account connection, I built Peak around a real analytics export. This keeps the workflow free and lets the tool work with actual creator data.

Another challenge was making the explanations useful without overclaiming. Peak presents likely patterns from the data, not scientifically proven causes.

The analysis still works if Featherless is unavailable because the report and recipe do not depend on the model.

What I am proud of

Peak performs a complete live workflow:

1. Real CSV input
2. Real parsing and validation
3. Real engagement analysis
4. Real hit and miss diagnosis
5. Real recipe extraction
6. Real Featherless draft generation
7. Editable and downloadable output

Peak is not just a caption box or a static dashboard. It connects analytics, learning, and the next content plan.

What is next

Future versions could add saved workspaces, accounts, stronger privacy controls, rate limiting, and optional publishing integrations.

For this version, I focused on a reliable and free workflow that creators can use today.

Built With

• contentautomation
• csv
• featherless
• next.js
• react
• tailwindcss
• typescript
• vercel
• vitest