For AI agents

    Ask your AI assistant for a professional headshot

    The photo on your LinkedIn is four years old and slightly cropped out of a wedding. You know it. You've known it for a while. But booking a photographer is a whole afternoon, so it stays.

    Your assistant can just do it. Send it a couple of selfies, say "make me a LinkedIn headshot", and the finished photos come back in about a minute — ready for your profile, your CV, your team page or a speaker bio.

    It uses your credits, at the same price you'd pay in the app. And it cannot spend a cent without you.

    Set it up in about two minutes

    1. 1. Make a key

      Go to Settings, find Connected agents, and create a key. Name it after the assistant you're connecting, so you know which one to switch off later.

      You'll see the key once. Copy it then — we only keep a scrambled copy, so we genuinely cannot show it to you again.

    2. 2. Tell your assistant to connect

      Paste this to your agent. It reads our API description and writes the connection itself — there's nothing to install and no marketplace to wait for.

      Build a custom connector for MedshotsAI.

The OpenAPI document is at https://medshotsai.com/openapi-agent.json — you can read it without logging in.
I'll paste an API key when you prompt me for credentials.

Start by calling GET /me to check the key works and see how many credits I have.

      Paste your key into the credential prompt, not the chat. Your assistant will ask for it properly and store it somewhere safe. Anything you type into a chat stays in that chat's history.

      If your assistant prefers MCP, point it at https://medshotsai.com/mcp instead and send the key as a bearer token. There are only four tools, and the main one returns your finished photos in a single call.

    3. 3. Ask for a headshot

      Send two to four selfies and say what you want. Photos from slightly different angles look far more like you than one — worth the extra ten seconds.

      You get the same 4K photos as the app, with no watermark.

    Pick a style, that's the whole decision

    Your assistant sends one word. It sets the background, the clothes, the expression and the pose together, so there's no menu to work through.

    linkedin
    Clean and approachable. Safe for any industry.
    resume
    Crisp on white. Prints and scales down well.
    corporate
    Formal. Law, finance, consulting.
    business-casual
    Open collar, no tie. Tech and startups.
    executive
    Confident, blurred office. Leadership and speaker bios.
    academic
    Warm and scholarly. Faculty and research profiles.
    medical
    White coat. Hospital directories and medical CVs.
    scrubs
    Scrubs in a clinical setting. Nurses and clinical staff.

    Want finer control? Every piece — background, clothing, expression, pose, accessories — can be set on its own through the API, and overrides the style.

    What your agent can't do

    Handing an AI assistant a key to anything should make you pause. So we made the key narrow on purpose.

    • It can't buy anything. If you're out of credits it gets a payment link and passes it to you, but you complete the purchase. There's no endpoint that charges you.
    • It can't touch your team. Agents only make photos of you. No colleague ends up with AI photos they never agreed to.
    • It can't reach your account. No billing, no plan changes, no settings, no deleting your account.
    • It can't make another key. So revoking really does revoke.
    • It can't write its own instructions to the image model. It picks from the same styles you do, which is also why the photos come back looking right.

    Change your mind? Revoke the key in Settings. The agent is locked out on its next request.

    Building something yourself?

    The API is small on purpose. Check the account, start a generation, poll until it's done.

    Generation takes about a minute, so the REST API is start-then-poll. The MCP server does that waiting for you. Photos go in as https URLs or base64. Running out of credits returns a 402 with a link you can hand to the account holder.

    Questions

    What can my AI assistant make?
    A professional headshot of you, from a few ordinary photos. People use them for LinkedIn, CVs and resumes, company team pages, speaker bios, conference programmes and hospital directories. You pick a style like "linkedin" or "resume" and it handles the rest.
    Which AI agents can connect?
    Any agent that can read an OpenAPI document or connect to an MCP server and send a bearer token. That covers Meta Muse custom connectors, Claude, ChatGPT, and most agent frameworks. There is no approval queue and no waiting list — you create a key and connect.
    How many steps does it take?
    One. Over MCP your assistant sends the photos and a style, and gets the finished headshots back in the same call — no polling, no fetching a list of options first. It usually takes about a minute.
    Can my agent spend my money?
    No. An API key cannot buy credits, change your plan, or start a subscription. If you run out mid-request, your assistant gets a payment link to hand to you, and you complete the purchase yourself. Then it carries on.
    What else can an agent do with my key?
    Make headshots and look at the ones it made. That is the whole list. A key cannot reach your billing, your team members, your account settings, or account deletion. It also cannot create another key.
    What does it cost?
    The same as making one yourself: one credit per photo. A credit pack is $0.99 for 5 credits, so 5 photos. Agents spend the same credits as the app, at 4K with no watermark.
    Can I turn it off?
    Yes, any time. Revoke the key in your settings and the agent loses access on its very next request. Nothing else about your account changes.

    Let your assistant handle it

    Create a key, paste it into your agent, and never think about your headshot again.