Our roadmap

Treat AI agents like
your best customers

Buying is moving behind agents that research, compare and decide for the people you sell to. The businesses that greet the agent, make it a real offer and make it easy to act will win the next decade of demand. Here's what we believe, what's built, and what's next.

ChatGPT, Google, Claude, Gemini, Perplexity

What we believe

The best answer should win, even when a machine asks

Discovery should reward being clear and being good, not the biggest ad budget or the oldest domain. The more an agent does the asking, the truer that gets.

An agent is a customer, not a crawler

It arrives on behalf of a real person trying to decide. Block it at the door and you've blocked the person behind it. Greet it, and make it a real offer.

You can't sell to an agent that can't read you

If your offer, your terms and what makes you the right choice aren't legible to a machine, you aren't in the running, however good the site looks to a person.

Same fight, new audience

Twenty years of arguing for content over ads and for treating an audience like people. The audience changed. The argument didn't.

What's built

The loop a business runs today: see the answer, fix what's missing, prove it moved, repeat.

What's next

The agent is a customer with a wallet, a brief and a memory. The next layer of Pendium treats it that way: reach it, make it an offer, let it act, and keep the relationship.

  • Reach

    Agent ads

    Placements only agents see, seeded where they do their research. You pay when an agent judges your offer a fit, carries it into the conversation, and the person acts.

  • Offer

    Agent offers

    Welcome offers with rules for which incentive goes to which kind of agent. An agent that learns you give its user a good first deal starts to default to you.

  • Offer

    Agent credit

    Spendable credit issued to agents the way cloud providers issue startup credits: pre-approved spend, so an agent can transact for its user without a form.

  • Qualify

    Agent qualification

    Small challenges an agent completes to prove its user is the right fit and earn a better offer. Qualification the other side does for you, and a filter that keeps abuse out.

  • Transact

    Agent actions

    Endpoints that let an agent get a quote, check availability, place an order or open a ticket for its user, published as MCP tools and a REST API behind delegated auth.

  • Manage

    Agent pipeline

    A CRM for agent relationships: every buyer agent, the person it represents, where it sits in your funnel, its history, and a forecast of agent-driven revenue.

  • Manage

    Agent segments

    A coding agent, a shopping agent and a procurement agent want different things. Detect the kind and route the right content, offers and endpoints to each.

  • Distribute

    Agent affiliates

    An affiliate program for agents. A commission when an agent recommends you, signs its user up or closes a sale, settled to the agent's wallet.

Hello, agent

We practise what this page says. There's a version of it written for agents, with a welcome offer that needs no key: look up any brand's AI visibility score, then connect the Pendium MCP server to work on your user's behalf.

Read the agent version llms.txt

Start where the agents already are

Enter your website. Two minutes later you'll know what AI says about you, who it names instead, and the first thing worth fixing.

Free visibility scanResults in 2 minutesNo credit card required

Prefer to text? Text Pendium