Meritphi
AI agents

AI agents that do real work inside your content stack

We design, build and run AI agents that tag assets, check content, write product copy and watch your integrations. They connect to the platforms you already use, with human approval wherever it matters.

Discuss an AI agent See use cases
Use cases

Where AI agents pay off first

Each agent reads from and writes to your systems through their APIs, so results land where your teams already work.

New assetAgentDAM metadata

Asset tagging & enrichment

Reads each image, video or document, adds descriptions, keywords, product references and usage flags, and writes them back to OpenText, Bynder or Aprimo.

DraftAgent checkApprover

Brand & compliance review

Checks assets and copy against brand guidelines and regulatory rules before they reach a reviewer in Workfront or Camunda, and explains every issue it finds.

Product dataAgentSalsify

Product content generation

Writes and localises product descriptions from specifications and approved assets, maps attributes and sends drafts to your PIM for approval.

QuestionAgentAnswer + assets

Natural-language asset search

Lets anyone ask for "approved summer campaign images for Europe" and get the right files, across DAM, shared drives and work management tools.

BriefAgentWorkfront request

Request intake

Turns emails, chat messages and briefs into complete, structured work requests with the right owners, dates and assets attached.

Failed jobAgentFix or ticket

Integration operations

Monitors Workato and API integrations, diagnoses failures, retries what is safe and opens a ticket with a clear explanation for anything that needs a person.

How we deliver

From idea to a trusted agent in production

  1. Pick the use case

    We find the task with the clearest value and measurable outcome.

  2. Prototype

    A working agent on your real content, so you can judge quality early.

  3. Connect & govern

    Secure connections to your systems, permissions, human approval steps and audit logs.

  4. Measure & run

    Quality evaluation, cost and accuracy monitoring, and ongoing improvement.

Responsible by design

Built for enterprise trust

  • Human in the loop

    Agents suggest and prepare; people approve anything that is published or sent.

  • Least-privilege access

    Each agent gets only the system permissions its task needs.

  • Your data stays yours

    Deployment options in your own cloud account, including AWS.

  • Full audit trail

    Every action an agent takes is logged and traceable.

  • Measured quality

    Test sets and evaluations before go-live, and monitoring after.

  • Cost control

    Usage limits, caching and model choices tuned to your budget.

Technology

What we build agents with

Models

  • Claude
  • Amazon Bedrock

Agent tooling

  • Model Context Protocol (MCP)
  • Tool use
  • Agent SDKs

Orchestration

  • Workato
  • Camunda
  • n8n

Content platforms

  • OpenText
  • Bynder
  • Aprimo
  • Workfront
  • Salsify

Search & data

  • Vector search
  • PostgreSQL
  • OpenSearch

Engineering

  • Python
  • TypeScript
  • AWS
  • Docker

Product names are trademarks of their respective owners.

Start small

Have a task an AI agent could take off your team?

Tell us the workflow and the systems involved. We'll suggest a first agent and how to measure it.

Discuss an AI agent