HERMES AGENT · VIDEO WORKFLOW GUIDE
Kling 4.0 is a good fit for short videos with clear shots, references, and motion requirements. Hermes organizes the brief, writes the shot plan, and checks the result. Kling creates the video.
01 / Design the workflow first
A practical flow is: the main Agent confirms the goal, a text model writes the shot table, an image model prepares character or product references, Kling creates the clips, and a person checks the brand, motion, captions, and product facts.
Even before Hermes lists a direct Kling connection, you can use Hermes to write the shot list, prompts, and review checklist. When Kling becomes callable, only the last step needs to change.
02 / Three real work scenarios
The owner gives Hermes the product page, target country, offer, banned claims, and three reference images. Hermes turns them into a shot list: opening product view, one benefit, a proof moment, and a clear call to action. Kling creates three short variants from the approved references. A person checks the price, logo, language, product shape, and rights before publishing.
A small studio first uses a cheaper image model to explore the character, costume, location, and lighting. After the style is approved, Hermes keeps a continuity sheet and sends Kling the reference board plus start, middle, and end keyframes for each shot. Kling handles motion; Hermes records which version belongs to which scene. A person checks identity, hands, faces, and story continuity.
Hermes breaks the original ad into timing, camera moves, dialogue, product moments, and music cues. It then creates a replacement list for each market and product. Kling generates the new clips from the approved references. A person checks that the product is not warped, the lip sync still matches, local claims are allowed, and the original footage and voices can be reused.
03 / Omni Reference or Keyframes?
They solve different problems. Do not rely on both controls in the same task; choose the one that matches what you need to control.
Use it when you need to keep several people, products, styles, or locations in view as references.
Use it when you need to control the opening, ending, and middle transitions of a shot.
A single generation is about 3 to 30 seconds. Longer stories should be split into shots and edited by a person.
Common landscape, portrait, and square formats are available. The exact highest setting depends on the product account.
Motion, identity, captions, logos, prices, and rights should not be approved by the model alone.
04 / How should you estimate the cost?
Kling 4.0 is currently sold through product plans and credits. The real cost changes with your account, shot length, resolution, and mode. Web credits are not the same as a Hermes API unit price, so test with a lower setting before producing a full batch.
View Kling pricing ↗05 / Hermes API
The public Hermes model directory does not list a Kling 4.0 model ID yet, so this page has no “call Kling directly” button. You can still use Hermes text and image models to prepare the workflow, then connect the final step when Kling is officially listed as a callable model.
Read the Hermes API docs ↗06 / FAQ
Not as one generation. Split a longer story into 3 to 30 second shots, then edit the clips together. This gives you more control.
Start with Omni Reference when identity and product consistency matter most. Start with Keyframes when the start, end, and transitions matter most.
At the time of this review, Kling 4.0 has no model ID in the public Hermes directory. This page does not present it as an available Hermes API route.
Someone who knows the brand and the product should review it. Check shape changes, captions, prices, and music or image rights.
Official sources
HERMES AGENT
You can use Hermes today to plan the script, references, and review checklist. Kling 4.0 creates the video; direct Hermes access depends on the public model directory.
View the Hermes model directory →