Agency Pitch Mockups

Agency pitch mockups

Rapid visuals for a new-business pitch deck: campaign hero, product tiles, key-visual explorations, quote slides — all client-branded, all clearly-labeled-as-mockup, all iterable in under an hour before the pitch.

Brief + locked direction → full pitch-deck asset set. Speed first (pitch windows are brutal), reusability second (if you win, these become production inputs), clarity third (nobody should confuse a mockup for shipped work).

When to Use

• Net-new business pitch, 48-hour turnaround
• Mid-pitch concept swap — client leaned toward direction 3, need assets tonight
• Multiple concepts per pitch (A/B/C) — each needs its own hero + 3 tiles
• Internal review deck for a signed client's next campaign
• Pre-approval creative exploration before any photo/video production spend

Do not use for finished deliverables — pitch mockups are intentionally aspirational and mockup-flagged. For final assets after win, use agency-multi-brand-pack or project-specific production workflows.

Prerequisites

Ask the user (one message):

1. Client slug + brand file — path to clients//brand.md (output of agency-brand-scoping). If missing, run brand-scoping first.
2. Pitch concept(s) — one tagline or idea per concept (e.g. "Banking that waits for you")
3. Deliverable breakdown — how many heroes, tiles, quote slides per concept? Default: 1 hero + 3 tiles + 2 quote slides
4. Pitch date — drives whether we run cheap drafts or go straight to finals
5. Confidentiality — NDA status; affects Drive folder + any watermark choice
6. Mockup labeling — visible "MOCKUP" watermark, subtle footer, or none? (default: subtle footer for external pitches)

If the user hasn't locked a direction yet, stop and route them to agency-brand-scoping first. Don't pitch on guesses.

How to Run

1. LOAD BRAND    → read clients//brand.md + brand-system.json
2. PLAN          → table of every asset with prompt + aspect + model
3. ESTIMATE      → gen-ai pricing on the full manifest (target /prompts.json` — when you win, production starts here.

## Pitfalls

- **Over-polished mockups** that the client expects you to deliver for pitch prices — add mockup footer + use draft models
- **Skipping ` — palette drift mid-manifest, deck looks incoherent
- **Cross-client prompt contamination** — don't copy prompts from another client's repo without stripping brand specifics
- **Using real faces / real brand marks** — rights issues; use descriptors ("a founder in their 30s") and generic marks
- **Missing rights documentation** — if the pitch wins, you'll need to reproduce these with clean provenance; save `results.json` per run
- **No `gen-ai pricing` before a 50-variant concept explosion** — pitch budgets don't absorb $40 surprises

## Verification

Run `gen-ai whoami` to confirm authentication, then re-run the failed command with `--debug`.

## Step 1: Plan the asset set

Present a table before spending:

| # | Asset | Concept | Model (draft → final) | Aspect | Notes |
|---|---|---|---|---|---|
| 1 | Hero | "Banking that waits" | `recraftv4` → `flux-2-pro` | 16:9 | Full-bleed, tagline overlay |
| 2 | Tile A | Product focus | `recraftv4` | 1:1 | Clean product shot |
| 3 | Tile B | Audience focus | `recraftv4` | 1:1 | Founder at desk, moody |
| 4 | Tile C | Benefit focus | `recraftv4` | 1:1 | Abstract benefit icon |
| 5 | Quote slide 1 | Customer voice | `ideogram-v3` | 16:9 | Readable headline type |
| 6 | Quote slide 2 | Founder voice | `ideogram-v3` | 16:9 | Readable headline type |

Estimate the manifest. If under $5, proceed. If over, cut the draft resolution or concept count.

## Step 2: Draft manifest + run

```bash
CLIENT="acme-fintech"
CONCEPT="waits-for-you"
OUT="clients/$CLIENT/pitches/$CONCEPT/drafts"
mkdir -p "$OUT"

cat > /tmp/pitch-$CLIENT-$CONCEPT.json <<EOF
{
  "defaults": {
    "model": "recraftv4"
  },
  "jobs": [
    { "id": "hero",       "prompt": "$CLIENT hero, concept: banking that waits for you — editorial, cinematic, headline overlay area left-third", "aspectRatio": "16:9" },
    { "id": "tile-01",    "prompt": "$CLIENT product tile — clean product render, brand palette",           "aspectRatio": "1:1" },
    { "id": "tile-02",    "prompt": "$CLIENT audience tile — founder at desk, moody lighting, documentary", "aspectRatio": "1:1" },
    { "id": "tile-03",    "prompt": "$CLIENT benefit tile — abstract waiting metaphor, minimal composition", "aspectRatio": "1:1" },
    { "id": "quote-01",   "prompt": "quote slide, large editorial headline 'Finally a bank that moves like I do.', brand palette, minimal background", "aspectRatio": "16:9", "model": "ideogram-v3" },
    { "id": "quote-02",   "prompt": "quote slide, large editorial headline 'Built for founders who don't wait.', brand palette, minimal background",   "aspectRatio": "16:9", "model": "ideogram-v3" }
  ]
}
EOF

gen-ai batch run /tmp/pitch-$CLIENT-$CONCEPT.json -c 4 -o "$OUT"

Review internally. For approved concepts, re-run the subset with model: flux-2-pro and output to pitches/$CONCEPT/finals/.

Cost & time

| Phase | Spend | Time | |---|---|---| | Draft pass (6 assets, recraftv4) | ~$1-2 | 10-15 min | | Final upgrade (2-3 winners to flux-2-pro) | ~$2-4 | 15-20 min | | Total pitch kit | ~$5 | ~45 min end-to-end |

Compare to production-phase cost after win: single campaign set with final photography-grade generations runs $20-100+.

See also

• workflows/agency-brand-scoping/ — lock direction + brand.md before pitching
• workflows/agency-multi-brand-pack/ — retainer production across clients
• workflows/agency-client-handoff/ — package final deliverables post-win
• gen-ai-use — if the pitch needs a sizzle reel
• gen-ai-workflows.md §9 (Pitch mockups) — source recipe
• gen-ai-batch.md — manifest + concurrency tuning

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: PicsArt
• Source: PicsArt/gen-ai-skills
• License: MIT

Install and usage instructions live in the source repository linked above.