What Is Docket? AI Agents for Marketing Teams

Docket is a B2B software company that builds AI agents for marketing teams. Docket’s Demand Agent is an autonomous agent that chats with website visitors, answers product questions, qualifies intent, and books sales meetings.

Today, the live member of that team is the website agent. It talks with visitors on your site, answers their questions from content you approve, qualifies them, and books meetings with the right rep. Then it hands that rep the full conversation in your CRM.

In short, Docket turns website traffic that marketing has already paid for into qualified pipeline.

Docket at a glance

What does Docket do?

Docket’s website agent turns website visitors into booked, qualified meetings, without waiting for a form fill or a callback.

Here’s a hypothetical. A VP of operations at a regional logistics company clicks a paid search ad and lands on your integrations page. She wants to know two things: whether you sync with Dynamics 365, and what onboarding looks like.

On a form-based site, she fills in her details and waits for an SDR to reply. With Docket, she asks the website agent directly, in her own words. The agent answers from your approved docs, asks a couple of qualifying questions about team size and timeline, and offers her a meeting with the rep who owns her territory. She books it without leaving the page.

The rep gets more than a name and an email. The CRM record shows what she asked, what she needs, how she answered the qualifying questions, and the next step she agreed to.

That’s an Agent Qualified Lead (AQL): a buyer who has articulated their needs, demonstrated fit and agreed to a next step. Docket coined the term. An MQL guesses intent. An AQL confirms it. More on what makes a lead an AQL.

How does Docket’s website agent work?

The website agent starts conversations, remembers visitors, answers from approved content, qualifies buyers, books meetings and writes it all back to your CRM.

It starts the conversation at the right moment

Contextual Callouts open a conversation based on the page a visitor is on and how they’re behaving, so the agent doesn’t sit in the corner waiting for a click. Visitors can talk or type, in multiple languages, and switch between voice and text mid-conversation (how agent modes work).

The agent can also appear as an avatar. To see one, talk to Aura on docket.io: it’s Docket’s own website agent, running as an avatar. Text-only mode is there for pages where a lighter experience fits better. Avatar options vary by plan (see plans).

It knows who it’s talking to

Buyer Intelligence is what the website agent knows about a visitor before they type a word: CRM history, company details and live intent signals (how Buyer Intelligence works).

The Buyer Context Graph is where Docket keeps that knowledge: one continuously updated profile of each visitor and their company, combining first-party data, third-party signals and conversation history. Buyer Intelligence is what the agent knows when a visitor arrives. The Buyer Context Graph is where it keeps that knowledge, and it grows after every visit.

So when someone from the same account comes back next week, or a colleague shows up with a new question, the agent picks up where the last conversation ended instead of starting over. Anonymous visitors are recognized at the company level and known contacts at the person level. If a visitor declines consent, nothing is stored.

The Demand Agent answers only from content you approve

Answers come from the Sales Knowledge Lake™, the approved knowledge every Docket agent answers from. You choose the sources, such as your website, documents, decks, Google Drive, SharePoint or Intercom, and each agent uses only the subset selected for it.

When the agent doesn’t have an answer, it follows a set fallback instead of guessing. That’s what lets brand, legal and security teams approve exactly what it says (what a Sales Knowledge Lake is).

The Demand Agent qualifies buyers in plain language

You write qualifying questions in plain language, using BANT, MEDDIC or your own criteria, and the agent works them into the conversation. After each conversation, post-call analysis extracts the fields you define, such as a summary, pain points, qualification answers, budget and timeline, and can trigger actions and alerts.

The Demand Agent books the meeting with the right Sales rep

Routing works in two steps. The agent matches CRM ownership first. If there’s no match, it reads the conversation and your plain-language instructions to pick the right team, then books whichever rep is free, inside the conversation (how routing works).

It works alongside existing Chili Piper routers, syncs with Google and Outlook calendars, and falls back automatically when no rule matches. When a rep wants to step in live, the conversation is handed off in Slack or Teams.

The Demand Agent writes chat conversations back to your CRM

Every conversation syncs to Salesforce, HubSpot, Dynamics 365 or Marketo, so reps start the first call already briefed (example: the HubSpot integration; setup guides for Dynamics 365 and Marketo).

Buyers can explore your product live with Docket Showroom

Some buyers want to see the product before they’ll talk to sales. Docket Showroom, a paid add-on to the website agent, lets them explore your product live inside the conversation. They can take control or point to what they want to see, and pages you list stay hidden (Docket Showroom).

Showroom also works inside your product, with a co-browsing mode that keeps the agent on the right side of the screen (more on Showroom).

What results do Docket customers see?

Two Docket customers have published results from the website agent.

• Factors.AI: the website agent generated 23 meetings in two weeks, at over five times Factors.AI’s baseline conversion rate, and 77% of those meetings were booked outside business hours (testimonial on docket.io).
• Zenity: the agent held 206 conversations in a four-week window. In its first active period, it booked 10 meetings with no BDR at the top of the funnel, and conversations were written back to HubSpot (Zenity case study).

How does Docket help shorten the sales cycle?

Reps skip re-discovery, because every meeting arrives as an AQL with the buyer’s needs, fit and agreed next step already written down.

Docket reports across its customer base a 10% shorter sales cycle, along with 15% more qualified pipeline and a 12% higher win rate (docket.io).

The mechanism is simple. When a rep opens the CRM record before a first call, they already know what the buyer asked, which objections came up and what the buyer agreed to. The first call starts at evaluation instead of introductions (more ways to shorten the B2B sales cycle).

Is Docket a chatbot?

Docket is the evolution of AI chatbot for B2B websitesA scripted chatbot follows a decision tree. Docket’s website agent answers visitors in their own words, creating a real customer conversation, based on content and Sales knowledge you approve.

Legacy chatbots break when a buyer goes off-script with a specific integration question or a real objection, and push the buyer back to a form. The website agent answers from approved knowledge, follows a set fallback when it can’t, qualifies the buyer and books the meeting (chatbots vs AI agents).

How do marketing teams control the website agent?

Marketers change what the agent says in a sentence, choose what it can answer, and see exactly what it produced.

• Plain-language editing. Prompt Studio rewrites the agent’s instructions from a sentence you write, flags conflicts with existing rules, and keeps version history with rollback. It’s available on newer agents.
• Topics you choose. Keep the agent off support questions or any topic you decide, with a set fallback.
• One agent per need. Run separate website agents for each product line, language, segment or section of your site, each with its own knowledge and routing. Clone one to launch the next.
• Visibility. A dashboard of conversations, leads and meetings, an engaged-visitor table and full transcripts.
• Ask your conversations anything. Docket Demand MCP is a read-only connection from Claude, ChatGPT or any MCP client. Ask how your agents performed, which accounts engaged and what buyers asked. Access follows each person’s Docket permissions (setup guide).
• Answers for visitors’ AI assistants. With Docket WebMCP, built on a proposed web standard, a visitor’s browser-based AI assistant can ask the website agent product questions and get answers from your approved knowledge.

Docket is SOC 2 Type 1, SOC 2 Type 2, ISO 27001 and GDPR certified (security and trust). Audit reports, policies and controls are in the Docket Trust Center.

Who uses Docket?

B2B companies with a considered purchase, meaningful website traffic and a team that follows up on website leads.

Docket fits best when buyers need answers before they’ll agree to a call, when most of your traffic is buyers rather than existing users, and when SDRs or inside sales work website leads through routing rules.

Marketing usually leads. A VP of marketing or head of demand gen owns the outcome, a marketing ops manager runs the agents, and SDRs and AEs receive the meetings. SDRs still work those meetings; they just get qualified, briefed buyers instead of form fills.

Docket isn’t an ad platform or an outbound tool. It works the demand you’ve already created, starting with buyers who arrive on your website. It’s also not built for low-traffic sites, finite known markets or dealer-led sales.

Docket the company

Docket is a venture-backed company headquartered in Palo Alto, California.

In July 2024, Docket, then called DocketAI, announced a $15M Series A led by Mayfield and Foundation Capital (press release).

Docket builds in the space usually called agentic marketing: AI agents that take action in the funnel rather than only producing content (what agentic marketing means). Its focus is turning demand into pipeline wherever the buyer is, starting on the website.

Frequently asked questions

What is Docket?

Docket builds AI agents for marketing teams. Its live product, the website agent, answers website visitors from content you approve, qualifies them, books meetings with the right rep and writes the full conversation to your CRM.

Is Docket the same as DocketAI?

Yes. Docket used to go by DocketAI, with its website at docketai.com. The company now goes by Docket, at docket.io.

Is Docket an AI marketing tool?

Yes. Docket’s website agent is what many teams are looking for when they search for an AI marketing agent: an AI agent that engages buyers on your website, rather than a tool for writing content or running campaigns (what an AI marketing agent is).

Is Docket an AI SDR?

Yes. If you’re looking at AI SDRs for inbound, Docket’s website agent does that job: it answers questions, qualifies visitors and books meetings on your site. It also remembers each account across visits, lets buyers explore your product live with Docket Showroom, and hands reps the full conversation in their CRM. Docket isn’t an outbound tool (where Docket fits among AI SDRs).

Does Docket replace SDRs?

No. The website agent handles the first conversation, so every visitor gets an answer without waiting for a person. SDRs still work the meetings.

What does Docket integrate with?

Docket writes back to Salesforce, HubSpot, Dynamics 365 and Marketo, books through Google and Outlook calendars and existing Chili Piper routers, and hands off live conversations in Slack or Teams. It also integrates with Outreach. See the full list of apps and integrations.

How much does Docket cost?

Docket’s published plans start at $4K a month, billed annually, and include implementation by Docket’s customer success team. Docket Showroom is a paid add-on. See current plans on Docket’s pricing page.

How do you set up Docket?

Docket’s customer success team handles implementation. The usual steps are: connect your CRM, select the knowledge the agent answers from, set your qualifying questions and routing rules, then add Docket’s JavaScript snippet to your site (getting started guide).

Want to see the website agent in action? Talk to Aura on docket.io.