The ultimate guide to prompt engineering for AI-powered DAM search

Imagine a bustling retail warehouse full of products waiting to be sent to store shelves or online customers. The warehouse is vast, containing countless items arranged in seemingly endless rows. Without an organized system, quickly finding a specific product becomes a daunting task.

Now, picture an advanced robotic system that you can prompt to not only locate items with pinpoint accuracy but also suggest related products that match seasonal trends or promotional themes. In a digital sense, this is how prompt engineering and AI Visual Search work together in Canto’s digital asset management (DAM) system.

This blog will explore how the combination of these two technologies is transforming industries like retail and manufacturing, illustrating their potential to enhance workflows, improve compliance, and drive better outcomes.

What is prompt engineering?

Prompt engineering is about crafting precise, context-rich inputs to guide AI tools in generating the most accurate and relevant outputs. Prompt engineering serves as the bridge between human intent and machine understanding, ensuring AI delivers meaningful, actionable results.

For marketers, prompt engineering is like writing a creative brief for an ad campaign — the more detailed and tailored the instructions, the better the outcomes. By leveraging prompt engineering, marketers can fine-tune AI tools to align with their goals, whether it’s identifying the perfect asset in Canto’s DAM system, generating personalized ad copy, or refining customer insights.

As AI evolves, the ability to engineer effective prompts has become an essential skill, empowering marketers to harness AI’s full potential to create campaigns that resonate with their audience and drive measurable results.

What is prompt engineering in digital asset management?

Prompt engineering in Canto’s digital asset management platform enables users to craft precise queries that guide AI to deliver accurate, actionable results. It transforms asset discovery by refining searches, reducing time spent locating assets, uncovering reusable content, and streamlining workflows.

By aligning AI outputs with user intent, it minimizes duplication, lowers content creation costs, and enhances content strategies.

How prompt engineering and Canto AI Visual Search connect

Canto AI Visual Search leverages computer vision and machine learning to analyze and interpret visual content within Canto, allowing users to locate specific images or videos based on descriptive inputs.

Traditional searches in a DAM would rely exclusively on precise keyword and tag searches, but if your users don’t know what keywords to include in their searches or content isn’t tagged correctly, the results will be unhelpful:

When you are able to use AI and natural language, you can begin to refine your searches in order to get better results.

Let’s enable AI Visual Search and use, as an example, a typical search for an online retailer looking for images of: “model smiling with a white background:”

The search results are very general and can be improved using prompt engineering. Your new search query could be modified to: “asian female model smiling closeup with a white background:”

Now, you can use prompt engineering to refine your search even more and provide additional details: “asian female model with black hair smiling with red lipstick closeup with a white background:”

For organizations managing thousands of digital assets, prompt engineering and AI Visual Search alleviate the frustration of combing through current and historical assets lacking sufficient metadata and enable teams to find the right assets faster.

By utilizing Canto AI Visual Search, organizations can overcome content chaos, improve content collaboration, and reimage how they manage digital content entirely.

Prompt engineering best practices

One of the best frameworks used by prompt engineers is called SMART. While this acronym is used by prompt engineers to achieve the best outputs from LLMs, SMART can also be used to help you get the most out of your Canto AI Visual Search queries.

Here is how to create a SMART prompt:

Specific

Ensure the prompt is direct and unambiguous.

Instead of writing “construction workers,” be specific: “images of carpenters hammering nails on a roof.”

Measurable

Define the aspect ratio for images: landscape, portrait, square, widescreen.

Rather than writing “red shirts on a blue background,” add an aspect ratio for the images: “portrait images of a red blouse with a light blue background.”

Achievable

Keep the query realistic.

Remember that AI Visual Search is searching for assets exclusively from your Canto library. If the assets are not available, AI Visual Search will provide the most relevant examples according to your search terms.

Relevant

Focus on outputs pertinent to your needs.

If you need holiday shots, be specific about which holiday: “Christmas themed images of sweaters. Or, instead of “images of holiday sweaters,” you can specify: “lifestyle images of Christmas sweaters.”

Time-bound

Specify when the images or videos were taken: morning, afternoon, night, type of lighting, black and white, time period, etc.

An example search could be: “Lifestyle image of a family walking in a forest at sunrise in the morning.”

When utilizing Canto AI Visual Search, remember to search SMARTer, not harder.

Practical applications in DAM: Vertical-specific use cases

Manufacturing: Identifying compliance risks through visual search refinement

In manufacturing and construction, ensuring photography compliance with safety and regulatory standards is critical. Canto AI Visual Search capabilities help locate assets that may break compliance standards, such as images of workers without required safety gear (e.g., helmets, gloves, or goggles). Prompt engineering and search refinement enhance these capabilities significantly.

Refining searches based on safety equipment

By developing targeted prompts (e.g., “construction worker wearing a baseball cap with a nail gun daytime closeup side shot”), prompt engineering allows DAM users to find assets that demonstrate a lack of compliance. In the image below, we can see examples of how refining our searches with prompt engineering can help find non-compliant media:

Tagging historical assets for OSHA compliance

AI Visual Search can be used to retrospectively tag historical assets with OSHA compliance tags. For instance, images can be marked with tags like “OSHA compliant” if all required safety gear is visible, creating an organized repository of compliant images for future audits or training:

Identifying and tagging OSHA violations in historical assets

Similarly, AI Visual Search can help detect and tag images that display OSHA violations — such as “worker wearing a baseball cap nailing a wood wall.” These images can be set to a Restricted Status in Canto, which will remove visibility for all users. Additionally, digital rights can be embedded into the assets, thus ensuring compliance at scale:

Cost and resource efficiency

By automating asset searches and compliance checks, AI Visual Search reduces the need for extensive manual audits. This cost-effective solution minimizes the time and labor needed to maintain compliance, supporting lean operations without sacrificing thoroughness.

Retail: Enhancing product discoverability and brand compliance

In the retail industry, DAM systems play a crucial role in organizing and distributing product imagery and promotional content. Retailers rely on efficient brand management processes and the ability to quickly retrieve on-brand content to keep pace with seasonal trends, promotions, and new product launches. Canto AI Visual Search, combined with prompt engineering, can streamline these processes, ensuring that assets are easy to find, on-brand, and seasonally relevant.

Refining searches for brand compliance

Prompt engineering enables retail teams to refine searches for desired brand assets, such as logos, assets with color schemes, and specific product angles or settings. For example, a retail team could prompt Canto AI Visual Search to locate only those images where the brand logo is prominently displayed or filter out assets that don’t align with current branding guidelines:

Seasonal and trend-based search optimization

You can use prompt engineering to search seasonal assets with natural language like “lifestyle images of a diverse group of friends posing for the camera with christmas sweaters.” Contextual prompts that use the SMART framework reduce time spent searching for specific campaign materials. SMART contextual prompts also help marketing teams respond faster to market demands while aligning with upcoming seasonal or holiday promotions:

ROI and efficiency metrics

Implementing Canto AI Visual Search has been shown to improve asset retrieval speed, drastically reducing search time. Retail marketers can leverage these improvements to optimize campaign workflows, allowing teams to focus more on strategic planning and execution rather than searching for assets.

For example, you will notice in the search query above that “ifestyle” is misspelled — this is intentional because Canto AI Visual Search enables users to find assets through intuitive, conversational queries that have misspellings or inconsistent capitalization.

If you were to search using the exact same query with metadata exclusively, no results populate:

 

Once Canto AI Visual Search is enabled, you can rapidly discover all relevant images. It’s important to note that proper nouns and holidays do not need to be capitalized in an AI Visual Search query, as you can see from a query like “guy with a beard and glasses cheersing with red cups at a christmas party:”

Enhancing the customer experience

Let’s say you have a specific HEX code that needs to be available in a colored eyeshadow product. By including the hex code from the metadata as well as natural language, we can enhance the customer experience by quickly delivering on-brand and relevant digital assets:

Canto AI Visual Search is the future of DAM discovery

If you are seeking to streamline your content workflows, maximize content value, and increase discoverability for specific use cases in your content library, Canto’s DAM platform powered by AI Visual Search is your answer.

Canto is the leader in digital asset management, powering top brands to organize, find, collaborate, and distribute digital assets faster than ever before. When you combine the power of AI Visual Search with effective prompt engineering practices, teams gain increased efficiency, marketing and sales campaigns become more effective, and organizations can scale content management easily.

In addition to AI Visual Search, Canto offers everything you need to transform your content workflows:

• Metadata management tools: Smart Tags and text recognition organize content with metadata for you
• Facial recognition: Find images featuring specific people and add metadata quickly
• Canto PIM: Product-driven organizations can finally unite product information and digital assets in one platform
• Media Delivery Cloud: Publish images to storefronts, websites, and platforms automatically in the correct sizes and ratios from Canto
• Video Enhancements: Discover video clips instantly, add timestamps and tags, generate captions, and embed on sites directly from Canto
• Collaboration tools: Streamline content production and approval using Workflows, Workspaces, comments, and annotations
• Style Guides: Organize all brand elements in one place to align teams and channels around your brand identity
• Brand management tools: Control your content with approval statuses, expiration dates, and digital rights management tools

Ready to discover the power of Canto for yourself? Book a personalized demo and find out how AI Visual Search could benefit your brand today.

Prompt engineering frequently asked questions

What is the key to crafting effective prompts?

The key to crafting effective prompts lies in providing details and following the SMART framework: ensure prompts are Specific, Measurable, Achievable, Relevant, and Time-bound. The SMART prompts approach helps guide Canto AI Visual Search to deliver precise and actionable results tailored to your needs.