Skip to content

Context Manager

The customer context every agent can trust

Customer data, activation memory and business rules, unified in your warehouse and read by Dan, your LLMs and your agents. Dan proposes your definitions, and you accept the ones that fit.

Your favorite brands prefer DinMo

  • Nexity
  • Galeries Lafayette
  • Interflora
  • TUI
  • Manor
  • K-Way
  • celio
  • Lafuma
  • Click&Boat
  • Krüger
  • Respire
  • Bayard
  • Premier Vet Alliance
  • L’Équipe
  • Libération
  • The Bradery
  • Bpifrance
  • Karavel
  • Ducasse Paris
  • Conforama Suisse
  • Leetchi
  • Partenamut

Why it matters now

Tomorrow’s marketing teams up with agents

Agents take on the execution, your team keeps the direction. Before an agent acts, it needs to know your customers and your rules.

  1. Yesterday

    Tools your team operates

    Context lives in your team

  2. Today

    Assistants that answer

    Context lives in each prompt

  3. Tomorrow

    Agents act, people steer

    Context has to live in one shared place

Tomorrow, concretely

Four agents at work on the same customers

People steer

They set the goals and approve what matters.

Agents act

Each one supports someone on your team.

  • Retention lead

    Retention agent

    Spots customers drifting away and drafts the win-back plan.

    Is Claire still active?

  • CRM manager

    Next best action agent

    Proposes the next move for each customer.

    What did she already get?

  • Service advisor

    Service agent

    Briefs the advisor live, next best action included.

    What is she worth?

  • E-commerce lead

    Site chatbot

    Knows who it is talking to, hands over when it should.

    What may I tell her?

Underneath, the same three questions

  1. 01 · Customer data

    Who is this customer?

  2. 02 · Activation memory

    What have we already done with them?

  3. 03 · Business rules

    What are we allowed to do?

When each agent answers them on its own, you get as many answers as agents.

The problem

The problem

With DinMo, every agent answers on your rules Without context, every agent guesses

With DinMo context, the agent finds 3,380 high-value customers to win back, sends the message to Claire M. by email only, hides her phone number and waits for Camille D. to approve. Her unified profile, your definitions and your rules light up under it.

Your definitions, your consent rules and one shared truth, applied on every request.

  • Benefit: Your definitions. “Active customer” means an order in the last 90 days, for every agent.
  • Benefit: Scoped access. Phone masked, opt-outs excluded, every change waits for Camille D.’s review.
  • Benefit: One truth. Dan, your agents and your LLMs read the same churn risk.

Without context, the agent reads raw tables and answers 129,600 customers: it counted every account that logged in this year, exposed Claire M.’s phone number and sent the audience without approval.

It invents your definitions, reads fields it should never see, and answers differently from the agent next door.

  • Risk: Wrong answers. Every account that logged in this year counts as active.
  • Risk: Leaks. The whole customer table, phone numbers and opt-outs included.
  • Risk: Drift. Three agents, three definitions of churn, three numbers.

What’s inside

Three layers. One truth.

DinMo turns your warehouse into one customer context. Each layer is kept in your cloud and read in place.

  1. 01Customer data

    One profile per customer, with orders, events, predicted value and churn risk attached.

  2. 02Activation memory

    The audiences, journeys and syncs DinMo runs, with their results, so every reader knows what was already done.

  3. 03Business rules

    Your definitions, consent rules and playbooks, written once in the Context Manager and read on every request.

Every change Dan prepares waits for your review before it runs. Every read by any agent stays within that agent’s scope.

Let Dan define your business

The definitions your industry needs, written with you

Start an analysis, and Dan studies your workspace for as long as it needs: your models, your fields, how your customers buy and subscribe. It comes back with the definitions your industry relies on, each with its formula and the evidence behind it. You review them, accept the ones that fit, and every answer, audience and prediction uses them from then on.

Approval: Promote 6 definitions to the workspace?

  • Monetary valueWhat counts as revenue: paid orders, net of refunds, in one currency.
  • Customer valueWho your valuable customers are, by spend, frequency or predicted value.
  • Subscription start and endWhen a subscription begins and when it ends, so churn means one thing everywhere.
  • It keeps learningWhen a request uses a word with no definition, Dan asks once and proposes to save your answer.

Nothing is shared until you accept it. Every definition keeps its versions.

Meet Dan

Who reads it

One context. Every team, every agent.

The same answer, on the same definitions, whoever asks.

Dan

Your marketing teammate. Asks in plain words, waits for your approval.

Your LLMs

Claude or ChatGPT read the context through the DinMo MCP plugin, within your scope.

Your agents

Your own agents call DinMo through the same MCP plugin, one scope per client.

Your teams and your website

Service, sales, editorial or product teams get the answer before the call, and your website reads it through the Personalization API. No data ticket.

How you build it

Your company context, built the way you work

Write it as a team, keep personal notes for yourself, import the definitions you already have, or start from a first draft.

Collaborative & personal

Team context, and yours

Your teams build the company context together. Each user adds personal context, like formats or habits, read only in their own conversations.

Many types of context

Your business, described

Business definitions, company facts, rules, periods, channels, playbooks, and whatever else your business runs on. The knowledge that makes an agent answer like your best analyst.

Semantic layer sync

In sync with your metrics

Metrics already defined in dbt, your warehouse or a spreadsheet? Connect the source: DinMo imports the definitions and keeps them in sync.

Drafted by Dan

Pre-filled by Dan

Dan analyzes your models, segments and fields, factors in your industry, then proposes the definitions worth sharing. Your team reviews and approves.

FAQ

Questions from data teams

Is DinMo a CDP or a semantic layer?

DinMo brings both capabilities together in one platform. For marketers, it’s an agentic CDP that helps them move faster every day. For AI agents, it provides a customer context layer with unified profiles, business definitions, rules, and governed access. If you already have a semantic layer, DinMo’s open, composable architecture lets it stay in sync with your existing definitions, so you maintain a single source of truth.

Does our customer data leave our warehouse?

Your customer tables stay in your warehouse and DinMo queries them in place. Data leaves only when you activate it to a tool, or when an authorized agent reads the fields its scope allows. If you choose DinMo Hosting, you can have your warehouse hosted in either the EU or the US.

Which agents and LLMs can connect?

Dan, and through the DinMo MCP plugin your own agents and assistants such as Claude, ChatGPT, Gemini or Mistral. Your website and apps read the same context through the Personalization API. Each client gets its own scope.

How do you stop an agent from seeing what it shouldn’t?

Every client has a scope on fields, models and actions, agents act within the permissions of the person they work for, and sensitive fields are masked by policy. Every change Dan prepares waits for your review before it runs, and every read and action is logged.

Context Manager

Give every agent the whole picture

See how DinMo turns your warehouse into governed customer context, on your own data.