Turn hard-to-use video into reviewable work
Composer helps media teams find moments, build first cuts, and package archive footage. Sentinel helps operators verify incidents and assemble evidence. Both keep people in control.
Partners and customers we work with
Why Pixel ML
We package real workflows, not generic AI promises. Here is why teams trust us with high-trust archive and operational video work.
Customer work with clear scope
Case studies explain the workflow, deployment model, and evidence behind each story.
- ◆Archive-to-output and investigation workflows in customer-controlled environments
- ◆Claims kept to the workflow and deployment details that can be shared
- ◆Media and operational use cases that show reuse, not just demos
Your AWS or ours
Some teams want managed delivery. Others want customer-cloud deployment for tighter control.
- ◆Customer cloud deployments with full control
- ◆Managed delivery where speed matters more than infrastructure ownership
- ◆Case studies distinguish the deployment model instead of blending them together
AWS ecosystem
Listed on AWS Marketplace with deployment options for customer-controlled environments.
- ◆Use AWS Marketplace when procurement needs that path
- ◆Deployments using AWS-native building blocks where they fit the workflow
- ◆NVIDIA Inception · Intel Liftoff · MMA Global award
How delivery works
Start with one painful workflow. Expand only after users trust the results.
- ◆4-8 week pilot projects
- ◆Human review on high-stakes or judgment-heavy decisions
- ◆Scale after workflow fit and production quality are proven
The team
Bootstrapped operators building product and customer deployments in parallel.
- ◆Patent pending for agent orchestration
- ◆AWS Marketplace partner
- ◆MMA Impact Vietnam 2024 — #1 Interactive Award
Useful links:
How Composer gets adopted
Start with one archive or editorial workflow, prove it with users, then expand. Deploy in customer cloud when control matters.
Plan, search
One workflow first
We start with one painful archive workflow: search, review, summarize, clip, and package useful outputs. No stack replacement required.
Proof before scale
Public cases keep the story at workflow level. Detailed architecture, metrics, and limits belong in an active evaluation.
Start with the video work that hurts
We do not ship a generic video AI layer. We build narrow workflows around search, review, clipping, evidence, and approval.
Selected work
Aviation video retrieval, newsroom operations in customer cloud, and enterprise content automation.
Customer results
Results from teams using Pixel ML systems in production.
We built a 16-agent content engine on AgenticFlow that researches, outlines, fact-checks, drafts, sends to humans for sign-off, and publishes. Article production went from about two weeks to 30 minutes.
BeaconSites
Digital marketing agency
Interactive AI image generation at CSE Job Fair 2025 — attendees created personalized AI art on the spot. Cho Tot publicly highlighted the activation as part of making AI accessible to Vietnam's tech community.
Cho Tot
Carousell Group · Event activation
We built a 16-agent content engine on AgenticFlow that researches, outlines, fact-checks, drafts, sends to humans for sign-off, and publishes. Article production went from about two weeks to 30 minutes.
BeaconSites
Digital marketing agency
Interactive AI image generation at CSE Job Fair 2025 — attendees created personalized AI art on the spot. Cho Tot publicly highlighted the activation as part of making AI accessible to Vietnam's tech community.
Cho Tot
Carousell Group · Event activation
We built a 16-agent content engine on AgenticFlow that researches, outlines, fact-checks, drafts, sends to humans for sign-off, and publishes. Article production went from about two weeks to 30 minutes.
BeaconSites
Digital marketing agency
Interactive AI image generation at CSE Job Fair 2025 — attendees created personalized AI art on the spot. Cho Tot publicly highlighted the activation as part of making AI accessible to Vietnam's tech community.
Cho Tot
Carousell Group · Event activation
We built a 16-agent content engine on AgenticFlow that researches, outlines, fact-checks, drafts, sends to humans for sign-off, and publishes. Article production went from about two weeks to 30 minutes.
BeaconSites
Digital marketing agency
Interactive AI image generation at CSE Job Fair 2025 — attendees created personalized AI art on the spot. Cho Tot publicly highlighted the activation as part of making AI accessible to Vietnam's tech community.
Cho Tot
Carousell Group · Event activation
Pixel ML is part of Intel's Liftoff program for AI startups. Intel featured AgenticFlow in developer ecosystem coverage around agentic AI and multimodal workflows.
Intel Liftoff
Ecosystem program
Pixel ML is part of Intel's Liftoff program for AI startups. Intel featured AgenticFlow in developer ecosystem coverage around agentic AI and multimodal workflows.
Intel Liftoff
Ecosystem program
Pixel ML is part of Intel's Liftoff program for AI startups. Intel featured AgenticFlow in developer ecosystem coverage around agentic AI and multimodal workflows.
Intel Liftoff
Ecosystem program
Pixel ML is part of Intel's Liftoff program for AI startups. Intel featured AgenticFlow in developer ecosystem coverage around agentic AI and multimodal workflows.
Intel Liftoff
Ecosystem program
Enterprise AI engagements, scoped up front.
Pricing follows your rollout size, operating model, and support needs.
Workflow SprintsPopular
Best for teams validating one serious workflow before a bigger rollout.
Included in this engagement:
- Scoped archive, media, or investigation workflow
- Defined success criteria and deliverables
- Human review checkpoints
- Post-engagement review and next-step recommendations
Pilot Programs
Best for teams testing AI on a single use case before expanding.
Included in this engagement:
- Single use case design and scoping
- Pilot delivery with human review checkpoints
- Customer-team enablement and feedback loops
- Clear outcomes to decide the next milestone
Production Deployments
Best for teams starting small and scaling once outcomes are proven.
Included in this engagement:
- Customer-controlled AWS deployment options
- Release process, support model, and escalation paths
- Security-aligned delivery
- Operating documentation for internal teams
Articles
FAQ
Quick answers about what Pixel ML does and how we work.