shhots Shhots UGC Ads
Turn a product and an audience into a finished UGC video ad: hooks, script, visual treatment, and generation through the Shhots AI MCP server.
Input: a product (photo, public image URL, or the active Shhots product), a target audience, and a list of allowed claims.
Output: 3 hook options, one structured script, one visual treatment formatted for shhots_create_product_video, and (with the MCP connected) the generated video.
Workflow
Step 1: Collect inputs in one consolidated question
Never ask piecemeal. Check the conversation for what is already known, then ask for the rest in a single message:
- Product: photo attached, a public image URL, or reuse the active Shhots product.
- Audience: who the ad targets, as specifically as the user can give.
- Allowed claims: the facts the ad may state. Explain briefly why: Shhots refuses invented benefits, certifications, ingredients, pricing, and performance claims, so the script can only use facts visible on the product or supplied by the user.
- Platform: determines the aspect ratio default (see table below).
- Language: Shhots generates UGC ads in 12 languages. Default to English unless told otherwise.
- Duration: 5, 10, or 15 seconds. Default 15 for a full hook-to-CTA arc.
If the user says "choose for me": UGC format, 15 seconds, 9:16, English.
| Platform | Ratio |
|---|---|
| TikTok, Reels, Stories, Shorts | 9:16 |
| Facebook/Instagram feed | 1:1 |
| YouTube pre-roll, landing pages | 16:9 |
Step 2: Write 3 hooks
Read references/hook-frameworks.md and pick the 3 frameworks that best fit the product and audience. Write one hook per framework, label each with its framework name, and keep every hook speakable in under 2 seconds. Present them and let the user pick, or pick the strongest yourself if they asked you to choose.
Step 3: Write the script
Structure for a 15-second ad (compress proportionally for 10s and 5s):
- 0-2s Hook: the chosen hook, spoken to camera.
- 2-7s Problem or context: the situation the audience recognizes.
- 7-12s Product moment: the product appears, in use, with the strongest allowed claim.
- 12-15s CTA: one clear action.
Pacing guidance: natural spoken delivery runs roughly 2.5 words per second, so a 15-second script is about 35-40 words, a 10-second script about 25, a 5-second script about 12. Write for the ear: short sentences, contractions, no marketing adjectives a real person would never say.
Step 4: Claim-safety pre-flight
Before generating, check every line of the script against the allowed claims list:
- No benefit, ingredient, certification, price, or performance claim outside the allowed list
- No superlatives presented as fact ("the best", "clinically proven") unless supplied by the user
- No implied medical, health, or financial outcomes
- CTA does not promise anything the user did not state
If a line fails, rewrite it using only allowed facts. Tell the user what you removed and why.
Step 5: Write the treatment
The treatment is the treatment parameter of shhots_create_product_video. Read references/treatment-templates.md and use the template for the chosen format. A complete treatment always covers, in order: setting, camera, lighting, actions, pacing, product moments, and CTA presentation. Use only visible or user-provided product facts.
Step 6: Generate (MCP connected) or hand off (not connected)
MCP connected: show the user the final script and treatment for approval, then call shhots_create_product_video with the approved script, treatment, videoType, durationSeconds, and aspectRatio. The tool displays a credit quote before generation; a video costs 551 to 852 credits depending on format and length. Most generations finish in under two minutes; product videos can take 1 to 3 minutes. Poll with shhots_get_generation if needed.
MCP not connected: deliver the script and treatment as a formatted brief the user can paste into the Shhots web app at app.shhots.ai. Mention that the MCP server (included on Shhots Pro and Scale plans, setup at shhots.ai/mcp) generates directly from chat.
Output format for the brief
ALWAYS use this exact template when delivering without generating:
# UGC Ad Brief: [product name]
**Format:** [videoType] · [duration]s · [ratio] · [language]
## Hook options
1. [framework]: "[hook]"
2. [framework]: "[hook]"
3. [framework]: "[hook]"
## Script (option [N])
[script with second markers]
## Treatment
[treatment]
## Claims used
[each claim and its source: "visible on product" or "user-supplied"]
Beyond UGC
The same workflow drives all 7 Shhots video formats: ugc, cinematic, animated, product_speaking, storefront, lifestyle, and unboxing_asmr. For non-UGC formats, the script becomes voiceover or on-screen text rather than creator speech; the treatment templates reference covers the differences.