imaga logo
UGC-style skincare video on a smartphone surrounded by cream jars, formulation notes, and lab tools

SMM Content Automation with AI

We developed an AI-powered SMM content system for a UK-based OEM cream manufacturer, replacing a manual workflow with a post-generation skill* and a node-based video pipeline.

*A Claude skill is a pre-built instruction package that teaches Claude to do one specific job the same way every time. Instead of prompting from scratch, you load the skill once and Claude follows its steps, rules, and references on every run.

United Kingdom
United Kingdom
Solution:
Workflow Automation
Services:
Business Analysis, Workflow Automation, AI Transformation
KPI:
Project Timeline: complete the project in 14 days.
Post production time: from 3 hours to 20 minutes per post.
Reels production cost: down 75–95% vs. market rates.

Client & Context

The client is a UK-based contract manufacturer. It produces creams for cosmetic brands under private label.

Buyers are brand owners, procurement leads, and R&D heads. They live on LinkedIn.

The in-house SMM team spent most of the week on manual content. No capacity for strategy.

Blank cosmetic bottle beside a smartphone mockup illustrating social media content and audience targeting

Goals

  1. 1
    Automate manual copywriting for social media posts.
  2. 2
    Replace paid UGC and animated videos with AI production.

Challenges & How We Overpowered Them

AI writes generic posts that do not serve the business.
The skill forces 8 validated steps. Audience and message come before writing.
The operator validates extraction, audience, message, and structure before a word is written.
AI hallucinates on formulation and regulatory claims.
The skill flags gaps — facts not in the source material are listed explicitly.
Source material comes from five in-house experts.
AI-generated content reads as AI-generated.
Copywriting rules ban forbidden constructions, clichés, and stamp phrases.
Mascot consistency breaks across videos.
A locked reference image is passed into every generation.
The pipeline standardizes prompts, lighting, and composition at the node level.

From AI Consulting to a Working Content System

AI Consulting Scope

A short consulting engagement came first. It mapped the client's marketing bottlenecks and identified SMM content as the single area where AI would pay back fastest. Everything below came out of that scoping.

Whiteboard with sales, ops, support, and SMM blocks, with SMM circled as the chosen automation focus

Post Generation Skill

Five in-house experts supplied raw material — project notes, formulation briefs, regulatory comments, factory walk-through transcripts. Before the skill, one post took 3 hours: interview, decode, write, edit.

The skill takes any input, structured or not, and runs it through 8 guided steps.

Post-writing AI skill structure with reference files for audience segments, copy rules, CTAs, and message framework

How the Skill Works

Extraction pulls facts and tension from raw material. The operator confirms audience, message, and structure from pre-built references. A plan gets validated.

Output: platform-specific versions for LinkedIn, Facebook, Instagram, and X, each with three alternative hooks and CTAs.

One post now takes 20 minutes end-to-end.

Before-and-after illustration comparing manual expert interviews with a 180-minute workflow to a 20-minute AI-assisted post process

Node Pipeline for Video

Two formats run on rotation: UGC-style talking-head clips and short mascot explainers. Both are expensive through traditional production.

The pipeline chains image models, video models, voice synthesis, music, and editing.

Node-based AI video pipeline showing content generation from text and visuals to video, voice, music, and final edit

UGC Videos

Images run through Nano Banana Pro, Nano Banana 2, or Recraft v4.

Video runs through Kling 2.5, Kling 3.0, or Sora 2.

Lipsync through ElevenLabs v3.

Final cut in CapCut.

Length: 15–30 seconds. Volume: 4 videos per month. One asset takes 10 minutes from prompt to export.

Laptop showing a node-based AI video workflow beside a cream jar on a minimal desk

Mascot Explainers

The mascot format explains formulation, compliance, and manufacturing topics in simple language. Same stack plus Suno v4.5 for original music.

Length: 15–30 seconds. Volume: 2 videos per month. One asset takes 30–60 minutes from prompt to export.

Smartphone displaying an AI-generated mascot skincare video next to a cosmetic cream jar

Compliance & Security

NDA-protected client data; no raw material stored outside client systems.
Human review on every generated asset before publication.
UK Cosmetic Regulation (UK SI 2013/1478)

Results

  1. 1
    Project delivered in 14 days from consulting to working pipeline.
  2. 2
    Post production: 3 h. → 20 min. 256 specialist hours freed per year.
  3. 3
    UGC video: 48 assets/year at £100–£300 market rate = £4,800–£14,400 saved.
Let's solve your challenge

See also

Scientific trial PDFs extraction
Scientific Trial PDFs Extraction
Self-hosted RAG chatbot
Self-Hosted RAG Chatbot for AKFA Holding

Request a project estimate

Up to 3 files, 5 MB each.
Your privacy is protected — we'll use your details only to reply to your inquiry. See our Privacy Policy.
What happens next
1
NDA. If checked, we'll send a standard NDA for e-signature right away.
2
Discovery. We'll reach out to clarify goals, constraints, and discuss potential AI use cases if they make sense.
3
Proposal. You'll get timeline and budget, plus key architectural options and risks.
We build for