I want to be upfront: we have a financial interest in you choosing to work with an agency. But we also have 320+ productions behind us, which means we've seen the full range of situations — brands that were well-served by DIY AI workflows, brands that made expensive brand consistency mistakes going it alone, and brands that came to us after burning time and budget on the wrong approach. What follows is what we'd actually tell a friend who asked us directly.
When DIY AI Visual Production Makes Sense
Social media content at high frequency is the clearest DIY use case. If you're running an Instagram account that needs 15-20 posts a month with product visuals, and the aesthetic bar is 'looks good in feed' rather than 'campaign-grade creative,' a skilled in-house person with Midjourney and Photoshop access can produce this reliably. The tools have improved enough that someone with a good eye and a few weeks of practice can get consistent, competent results.
Fast iterations and quick concepts are another legitimate DIY use case. Internal presentations, quick mood board approvals, trying out new product directions before committing to a full shoot — AI is genuinely fast for these and the quality bar is lower. You don't need agency precision for internal concept exploration.
Simple product isolation — removing backgrounds, placing products on clean surfaces — is now a commodity operation. Tools like Adobe Photoshop's Generative Fill, Remove.bg, and even built-in phone features handle this well enough that paying an agency for straightforward background removal doesn't make sense.
When Agency Work Makes More Sense
Brand campaigns are where the DIY approach starts to break down. Not because the individual images can't be good — they can be — but because a campaign requires a visual system to hold together across multiple deliverables, formats, and time. Colors need to match. The way light falls on a product needs to be consistent. The spatial relationships in compositions need to feel like they come from the same creative brain. Maintaining that across 50 or 100 images requires workflow discipline, quality control processes, and a trained eye at every checkpoint. Most in-house teams don't have the bandwidth for this.
Complex scene composition is another area where experience matters. Multiple products in a single image, specific background environments that need to feel art-directed rather than generated, product images that need to meet specific technical requirements for different placements — these require a level of control over the generation and post-processing pipeline that takes real time to develop.
If you're launching a new product and the visual identity is being established for the first time, get professional help. The decisions you make in those first images propagate everywhere — your website, your packaging, your ads, your social feed. Getting it wrong is expensive to fix. Getting it right the first time is worth paying for.
The Hidden Risk of DIY at Scale
The biggest problem we see from brands that have been doing AI visual production in-house for 6-12 months is not that any individual image is bad. It's that the overall visual identity has drifted without anyone noticing. The color of the product is slightly warmer in the Q3 content than in Q1. The product looks a bit smaller in some contexts. The background palette shifted gradually. None of it is catastrophic but all of it adds up to a brand that looks inconsistent to anyone paying attention — which your competitors and your more discerning customers are.
Cost Comparison
| Scenario | DIY AI | Agency + AI | Traditional Agency |
|---|---|---|---|
| Social content (20 posts/month) | Low — /month in tools | Medium — monthly retainer | High — per-deliverable |
| Product launch campaign (50 images) | Medium — time-intensive, risk of inconsistency | Medium-high — but controlled | High |
| Single hero product image | Low risk DIY if product already shot | Best quality-to-time ratio | Highest quality |
| A/B test creative variants (15 variants) | Feasible DIY with some training | Fast with quality control | Expensive at this volume |
| Brand visual system setup | High risk — get help | Recommended | Traditional still relevant |
| E-commerce catalog (100+ SKUs) | Feasible with right workflow | More efficient at scale | Rarely cost-effective now |
The Decision Framework
Ask yourself three questions. First: does this content need to last? Social posts have a short shelf life. Campaign hero images live on your website and in ads for months. The longer something lives, the more the quality matters. Second: how many people will see it? Internal presentation images don't need the same quality as homepage hero images. Third: is brand identity established? If the visual identity is still being defined, get professional help to define it. Once it's established, a capable in-house person can execute within it more independently.
- DIY: social content, fast iterations, internal concepts, simple product isolation, background removal
- Consider agency: campaign launches, new product visual identity, content at high volume with consistency requirements
- Use agency: brand campaigns, complex multi-product scenes, visual identity establishment, production where brand consistency is non-negotiable
- Always use agency: packaging design final artwork, any production with regulatory image requirements, brand identity creation
What an Agency Should Actually Provide
If you're going to work with an agency on AI visual production, the value should come from three things: creative direction (a trained eye that shapes what gets generated), process (workflows that maintain consistency across large volumes), and QC (someone whose entire job is noticing when something is off). If an agency is just running Midjourney prompts without these three things, you could do that yourself.
Frequently Asked Questions
How long does it take for an in-house person to get competent at AI visual production?
For social content with good tools: 4-6 weeks of serious practice. For campaign-quality work with consistency: 3-6 months. For managing complex scenes and brand-consistent visual systems: 12+ months or professional training.
What tools does a serious in-house AI visual workflow need?
At minimum: Midjourney or FLUX.1 access, Adobe Photoshop for compositing and retouching, a design tool (Figma or Adobe Express) for final layout. Budget roughly /month in tool subscriptions.
Can we start DIY and move to agency later if needed?
Yes, but be aware that visual drift during the DIY phase can create extra work when you bring in an agency to establish or restore consistency. Document your visual decisions as you go.
- Visual drift
- The gradual inconsistency in a brand's visual identity that accumulates when images are produced without strict reference to established brand standards.
- Visual system
- The set of rules, references, and standards that govern how a brand looks across all its visual outputs — colors, lighting style, composition conventions, typography.
- Creative direction
- The process of shaping the aesthetic decisions in a production — what gets made, how it looks, and whether it serves the brand — as distinct from the technical execution of generating images.
- QC (quality control)
- The review process that checks whether produced images meet the required standards before they're used in brand communications.
- SKU
- Stock Keeping Unit — a unique identifier for a product variant. E-commerce catalogs often have hundreds of SKUs, each needing its own product imagery.
If you're trying to work out whether DIY or a production partner makes more sense for your specific situation, we're happy to give you an honest answer — even if that answer is that you don't need us.