Production Templates

What Production Templates Do

A Production Template is the bridge between a discovered trend and a publish-ready asset. When the Discovery engine surfaces a high-scoring trend, the template takes that signal, combines it with your client's Brand Kit, and outputs structured content (copy, image prompts, video scene plans) that match the platform, tone, and visual identity exactly. Instead of writing prompts from scratch for every brief, templates encode the decisions your team has already made: which hook pattern to lead with, which color palette to reference, which call-to-action to close on. Every generation is repeatable, auditable, and brand-safe from the first run.

Template Types

Max Socials ships four built-in template categories. Each maps to a distinct set of platforms and output formats.

Text Post

Generates short-form copy for LinkedIn, X (Twitter), and blog short-form. The template controls character count, paragraph structure, hashtag density, and hook style (question, stat, contrarian statement). LinkedIn variants default to professional thought-leadership framing; X/Twitter variants default to punchy single-idea threads.

Image Carousel

Produces slide-by-slide copy and image prompts for Instagram and Facebook carousels. The template specifies slide count, per-slide word budget, the visual progression rule (e.g., light-to-dark palette gradient across slides), and the final-slide CTA structure. The Image Art Director picks the right generation pipeline (photorealistic, illustrative, or editorial) based on the brand's visual style profile.

Short Video

Outputs a scene plan for TikTok, Reels, and YouTube Shorts (15–60 seconds). Each scene plan includes hook text, B-roll description, caption overlay copy, and audio guidance. Remotion renders the scene plan into a composed video using the client's locked video template. Auto-captions are generated from the script in 50+ languages at render time.

Long-Form Blog

Generates a full article with a fixed intro/outro structure: hook paragraph, three to five body sections, and a closing CTA paragraph that mirrors the client's preferred conversion action. SEO title, meta description, and heading hierarchy are included in the output. Word count targets are configurable per client (default: 800–1,200 words).

Brand Kit Integration

Every template pulls from the client's Brand Kit at generation time. The Brand Kit contains four fields that templates read directly:

• Voice settings, descriptors like “authoritative,” “conversational,” or “witty” that the Content Writer uses to calibrate sentence length, vocabulary register, and sentence variety.
• Tone, the emotional temperature for the content: educational, inspirational, urgent, or playful. Tone can be overridden per template if the client runs distinct campaign types.
• Prohibited terms, a list of words, phrases, and competitor names the Brand Guardian will flag as violations. Any output containing a prohibited term is rejected before it reaches the editorial queue.
• Visual palette, primary, secondary, and accent hex values that flow into image prompts and video template color variables. The Image Art Director injects these into every generation call automatically.

Variables a Template Uses

At runtime, the Production engine resolves four variable slots before calling the AI:

• Trend hook, the headline angle from the Discovery engine (e.g., “Why [trending topic] is reshaping [industry]”). This becomes the opening line or scene-one hook.
• Brand voice, resolved from the Brand Kit voice settings; injected as a system-level instruction to the Content Writer.
• CTA preference, the client's configured conversion action: follow, click-link, download, or DM. Templates place the CTA in the position that performs best for that format (final slide for carousels, pinned comment for short video, last paragraph for blog).
• Visual style, drawn from the Brand Kit palette and any style-reference images uploaded to the client's account. Used to constrain image generation and video color grading.

Editing a Template

Navigate to Settings › Templates in the client account. Each template row shows its type, last-modified date, and the number of pieces generated against it. Click Edit to open the template editor, where you can modify the system prompt, adjust variable bindings, and set output constraints (max length, required sections, forbidden patterns).

Use the Preview button to generate a dry-run output against the current Brand Kit without publishing or consuming a generation credit. Preview outputs are stored for 24 hours under the Previews tab so you can compare iterations side by side.

To run an A/B experiment, duplicate an existing template and modify one variable (e.g., change the hook pattern from question to stat). Assign both templates to the same content bucket with a 50/50 traffic split. The Performance Analyst will surface which variant drives higher engagement after a minimum of 20 published pieces.

Custom AI Models (Agency Tier and Above)

Agency-tier clients can attach a fine-tuned model to any template instead of the platform default. Use fine-tuning when the client has a substantial corpus of high-performing past content (200+ examples) and needs consistent stylistic patterns the base model does not reliably reproduce, for example, a very specific sentence rhythm or a proprietary framework the brand teaches.

For most brand differentiation needs, prompt engineering is sufficient and faster to iterate. Fine-tuning adds a training cost and a 48-72 hour model preparation cycle. Start with prompt engineering; escalate to fine-tuning only when three or more prompt iterations still fail to capture the required style consistently.

Common Gotchas

• Too-narrow Brand Kit producing repetitive content. If the voice settings and tone are highly restrictive (e.g., “formal, data-driven, no humor”), the Content Writer has little creative latitude and will produce structurally similar pieces across different trends. Add two or three secondary tone modes to the Brand Kit (e.g., “storytelling allowed for case-study posts”) and bind them to specific template types to restore variety without breaking brand consistency.
• Missing prohibited-term lists causing Brand Guardian failures. The Brand Guardian only screens for terms it knows about. A client who has competitor names, legacy product names, or legally sensitive phrases that must never appear in copy will see those terms surface in outputs until the prohibited-terms list is populated. Populate it at account setup, not after the first rejection.