NEWMeet the ShareSmart AI Skilled Agent: early access is open ›Book a demoSupport
Share Smart AILog inSign up free
Help center

How can we help?

Search the answers below, or send us a message. A person replies within one business day.

Get started

Your first AI Skilled Agent in four steps

  1. Upload a documentStart with a real RFP, questionnaire, proposal or contract.
  2. Pick a playbookChoose a template, then edit the rules in plain language.
  3. Review its first answersSee what the agent would say before anyone else does.
  4. Share the agent, not the fileSend the link. Readers and their AI assistants can ask questions.
Want it done with you?

We will build your first agent

Bring a real document and we will set it up together.

Book a demo

FAQ

Quick answers

How do I create my first AI Skilled Agent?

Upload a document, pick a playbook, then review the agent’s first answers. Share the link when you are happy. The steps are above.

Which file types can I use?

PDF, Word, PowerPoint and Excel files, plus questionnaires.

Can I try it without a card?

Yes. The Starter plan is free and needs no card.

What happens to a document after I upload it?

The agent reads it, maps its sections, requirements, dates and obligations, and links every answer back to a page. You then review what it would say before anyone else sees it.

What is a playbook?

A set of plain-language rules: what the agent may share, what needs approval and who approves.

Why did the agent not answer a question?

The question probably falls outside your playbook. The agent holds it and asks you. You can approve, edit or decline, and add the rule to the playbook.

Can the other side use their own AI assistant?

Yes. Assistants can connect over A2A or MCP. A person can simply use the agent in a browser.

How does the agent decide which skilled agents to generate?

It reads what the document is for (a bid, a renewal, a security review or a negotiation) and who is reading it. Then it builds skilled agents from the matching playbook data. An RFP typically gets product, security, pricing and legal agents. A contract gets legal and compliance.

Can I control what the agent says and what it never says?

Yes. The playbook holds your approved positions and fallback language, topic blocks and redaction, approval thresholds and the named approvers. Anything outside it is held for a person.

How is this different from a chatbot on a PDF?

A chatbot answers from the text. An AI Skilled Agent also follows your playbook, generates skilled agents for the document’s intent, asks you to approve commitments, talks to other agents over A2A and MCP, and keeps an audit trail.

Can the same document give different answers to different readers?

Yes. The agent adjusts depth, tone and which skilled agents answer. A procurement reviewer gets price and terms first, a security reviewer gets controls and evidence, and legal gets clause-level answers with fallbacks.

How does the agent avoid making things up?

It answers from the document and the approved sources in your playbook, and each answer points to the page and version it came from. If nothing supports an answer, the agent says so and holds the question for you instead of guessing. You can always check the cited page.

What is A2A, in plain terms?

Agent-to-agent. It is the conversation between the recipient’s AI agent and yours. Their agent asks a question, yours answers, and the answer carries its source. Agents that support A2A can connect without a custom integration.

What is MCP and what can the other side’s agent reach?

MCP is how an agent requests specific data rather than asking a question. For example, it can request one section of your document. Your playbook decides whether that request is allowed, only the allowed section is returned, and the request is logged.

How does the recipient’s AI assistant connect?

The agent has a share link for people and an A2A address for other agents. Assistants that support A2A or MCP can use them directly. If theirs does not, the person can use the agent in a browser.

How do approvals work?

Commitments, pricing and policy questions go to a named approver before they are answered. You see one decision, with approve, edit or ask why. Until then the agent tells the other side the answer is pending, and the status changes to approved once you decide.

What happens when I update the document?

The agent answers from the new version and cites it. A new version of the same document does not count again toward your monthly volume. Readers and their agents can see when a pending answer becomes approved.

Does it connect to my other systems?

Agents connect to other agents over A2A and MCP. For connections to your own systems, such as custom connectors, see the Enterprise plan or book a demo and tell us what you use.

Do readers need an account?

No. Anyone with your link can read and ask questions. Your playbook still controls what they get.

Can I stop sharing a document?

Yes. Pause the agent or withdraw the document at any time.

Can I give different people different access?

Your playbook controls what each reader gets. You can also pause an agent or withdraw a document at any time, and the agent stops answering from it.

How does billing work?

Paid plans are priced by the number of documents you publish each month. You can pay annually or monthly. See the pricing page for plans.

How do I change or cancel my plan?

Contact support from this page. Changes take effect at the next billing date.

What counts as a document?

Each document you turn into an AI Skilled Agent counts once for the month. New versions of the same document do not count again.

Do readers and their AI agents count toward my volume?

No. People who read your AI Skilled Agent, and the AI agents that ask it questions, are free and unlimited. You pay for the documents you publish.

What happens if I go over my monthly volume?

We tell you before it happens and help you move to the next plan. Nothing is switched off without notice.

Why is monthly billing higher than annual?

Annual billing costs less per month. Use the toggle on the pricing page to compare.

Where can I find security information?

On the security page. For a full review, send us a request and we will reply within one business day.

What does the audit trail record?

Approvals, edits and answers: who decided what, and when. That gives you one version and one trail for each document.

Is my document used to train AI models?

We do not use your content to train models without your permission. The Privacy Policy explains what we collect and how it is used.

Can I get a security review before I start?

Yes. Send us your questionnaire or the questions you want answered, and a person on our team replies within one business day with documentation and answers.

No answers matched. Send us a message below and we will help.

Contact support

Talk to a person

Tell us what you were trying to do and what happened. We reply within one business day.

Selling to a team?

Book a demo and we will build an AI Skilled Agent from one of your documents.

Book a demo →

Questions about plans?

Compare plans and what each includes.

See pricing →

Send a message

Early-access teams get priority support.

This field is required.
Enter a valid work email.
Please choose one.
Please add a short message.

Message received

Thank you. We will reply to your work email within one business day.