agentura.

Agentura Blog

Author: The Agentura Team

  • Add an AI chat widget to your website

    Add an AI chat widget to your website

    Your website gets visitors with questions every day. A chat widget turns those quiet moments of hesitation into conversations — answered instantly, on your own site.

    Drop in one script, start answering

    Agentura gives you an embeddable website chat widget — a small embed script you add to your site. Once it’s in, visitors can chat with your agent right where they already are, without leaving the page or hunting for a contact form. The agent answers questions, helps people find what they need, and keeps the conversation moving toward a decision.

    It’s the difference between a visitor bouncing with an unanswered question and a visitor getting a clear answer in seconds.

    Make it look like yours

    The widget is themeable, so it fits the look of your site rather than feeling like a bolted-on box from somewhere else. Your visitors should feel like they’re talking to your business — the widget is built to blend in.

    The best chat widget doesn’t announce itself. It just feels like part of your site, ready the moment a visitor has a question.

    Safe by default for public visitors

    A widget on a public page talks to anyone who shows up, so strict security rules always apply to public visitors. That guardrail is on by default — the agent stays helpful within its bounds, so opening a chat to the world doesn’t mean opening up everything behind it.

    Every chat in one place — and it’s free to start

    Conversations from the widget appear in your dashboard alongside your other channels, so nothing gets lost and you always have the full picture of what visitors are asking. And there’s a low barrier to begin: website chat is included in the free tier, so you can put an agent on your site and see the conversations roll in before committing to anything.

    • Simple embed — one small script
    • Themeable to match your brand
    • Secure by default for public visitors
    • Unified dashboard for every conversation

    Add a chat widget to your website →

  • Your agent can answer the phone

    Your agent can answer the phone

    Not every customer wants to type. Some just want to talk — and now your agent can hold a real conversation in natural speech.

    An agent that speaks, not just writes

    Agentura supports voice agents through an ElevenLabs voice adapter, so your agent can talk with callers in natural speech. It listens, understands, and responds in a real conversation — the kind of back-and-forth a caller expects, rather than a rigid menu of press-one-for-this options. You pick the voice and the greeting it opens with, in the language your customers speak.

    For the customer, it feels like reaching someone who’s actually there and ready to help.

    The same knowledge, on the line

    The voice agent isn’t a separate, stripped-down system. It runs the very same agent as your text channels — the same SOUL, the same memory, knowledge files, and connectors. So whatever your agent can answer in chat, it can answer out loud — and whatever it can do through connectors, it can do for a caller too.

    A voice agent that only reads a script is an answering machine with better manners. The point is that it knows your business and can act on what the caller needs.

    Phone support and reception that picks up

    This is a natural fit for phone support and reception. Callers get answers immediately instead of waiting on hold, and routine calls get handled without tying up a person for every question. You set the hours your agent covers; outside them, it falls back to voicemail or a redirect, so nothing is dropped on the floor.

    Because the voice agent is the same agent you use everywhere else, it brings the same knowledge and connectors to the call — it works from the context it already has, not a separate cut-down brain. And when a call needs a person, your team can step in and take over the conversation.

    One agent, every way customers reach you

    Voice rounds out the picture: chat on your site, messages on WhatsApp and Telegram, and the phone — all the same agent underneath. Your customers choose how they want to reach you, and they get a consistent, capable response either way.

    It’s reception that’s always awake, phone support that doesn’t run out of patience, and a voice that actually knows your business. Telephony runs through ElevenLabs and Twilio, so calls reach your agent on a real line.

    Let your agent answer the phone →

  • An agent that takes WooCommerce orders

    An agent that takes WooCommerce orders

    “Do you have it in stock? How much is it?” Your agent shouldn’t just answer those questions — it should place the order while the customer is still asking.

    Live answers, not guesses

    The WooCommerce connector lets your agent read your catalog and check stock and prices in real time. When a customer asks what’s available or what something costs, the agent answers with live values straight from your store — not a stale copy that’s wrong by the time anyone reads it.

    That accuracy builds trust: the customer hears the real price, the real availability, and can act on it with confidence.

    Closing the loop on the order

    The interesting part is what happens next. The agent doesn’t stop at answering — it can place real orders in WooCommerce. The customer asks, the order is placed, and the conversation ends with something done rather than a to-do for your team.

    The customer asks, the order is placed. That’s the whole loop — closed inside a single conversation, with no copy-pasting into the back office afterward.

    Credentials the model never sees

    Connecting your store doesn’t mean handing your keys to the model. Your WooCommerce credentials are encrypted, and the model never sees them — it acts through the connector without ever touching the secrets behind it. The agent gets the ability to do the work without being trusted with the credentials that make it possible.

    Scoped to the right project

    Connectors are project-scoped, so a connection lives where it belongs rather than being available everywhere by default. The agents working on a given project see the connectors for that project — keeping access tidy and deliberate as you grow more teams and more stores.

    • Read the catalog and answer with live stock and prices
    • Place real orders from inside the conversation
    • Encrypted credentials the model never sees
    • Project-scoped access by design

    Build an agent that takes WooCommerce orders →

  • Issue invoices on the spot with Oblio & SmartBill

    Issue invoices on the spot with Oblio & SmartBill

    “Can I get an invoice for that?” should be the easiest question your business answers — not the one that ends up on someone’s afternoon to-do list.

    Real invoices, issued in the conversation

    Connectors for Oblio and SmartBill let your agent issue real invoices straight from your billing system. A customer requests an invoice, and the agent creates a real one — generated through your invoicing account, not a placeholder or a promise that the office will send it later. With Oblio, the agent can go one step further and submit the e-Factura to SPV, then read back the e-Factura status — so the Romanian e-invoicing leg is handled, not just the document.

    Because issuing an invoice is a deliberate act, both connectors ask for your confirmation before the invoice goes out. The agent does the work the moment it comes up; you give the OK. For a routine, repetitive task that nonetheless has to be done correctly every time, that’s exactly the kind of work an agent should take off your plate.

    Credentials stay sealed

    Issuing invoices means connecting to your billing accounts — and that’s done safely. Your credentials are encrypted and never exposed to the model. The agent acts through the connector to create the invoice without ever seeing the secrets behind your Oblio or SmartBill account.

    The customer asks for an invoice and, once you give the OK, receives a real one — issued and compliant — without anyone in your office digging through portals or copy-pasting figures.

    End to end: order, invoice, payment

    Invoicing rarely stands alone. Paired with the WooCommerce and PayPal connectors, your agent can run the whole sequence in a single conversation — from order to invoice to payment — chaining each connector in turn rather than handing the work off. The agent reads the order, issues the invoice, raises a payable PayPal invoice, and then checks back to confirm the payment landed.

    • Order read or placed via the WooCommerce connector
    • Invoice issued through Oblio or SmartBill
    • Payment requested via PayPal, then verified as paid

    Less back-office, more done

    The point of automatic invoicing isn’t just speed — it’s removing a recurring chore that’s easy to delay and easy to get wrong. When the invoice is issued the moment it’s requested — with your confirmation — your records stay current and your customers get what they need without waiting.

    That’s the difference between an agent that talks about your billing and one that actually does it.

    Automate invoices with Oblio & SmartBill →

  • Invoice & reconcile: PayPal & banking

    Invoice & reconcile: PayPal & banking

    An agent that can answer a billing question is helpful. An agent that can raise the invoice — and check whether the money landed — is a colleague.

    From conversation to invoice

    When a customer is ready to pay, you don’t want your agent handing them off to a form and hoping for the best. With the PayPal connector, your agent raises and sends a PayPal invoice inside the same conversation where the customer decided to buy — then reads back the transaction to confirm it cleared. It asks you to confirm before anything financial goes out, so nothing leaves on a guess. The flow stays unbroken, and the work that used to bounce between your support inbox and your finance tool happens in one place.

    Reading the financial picture

    Sending the invoice is only half of it. Banking connectors for Mercury and Revolut let an agent read your financial data, so it can answer questions grounded in what actually happened in your accounts rather than a stale spreadsheet. Ask whether an invoice cleared, what came in this week, or whether a customer’s transfer arrived — and get an answer from the source.

    Credentials are encrypted and injected at the network edge — never placed in the prompt the model sees. The agent uses the access without ever holding the keys.

    The right agent, the right account

    Money is exactly the kind of thing you don’t want crossing wires. Agentura connectors are project-scoped, so each agent reaches only the accounts that belong to its project. Your support agent for one brand never touches the finances of another, and a single misrouted request can’t reach the wrong account.

    • PayPal — raise and send invoices, and read transactions, without leaving the conversation.
    • Mercury & Revolut — read balances and transactions to answer with real numbers.
    • Encrypted everywhere — secrets live at the edge, not in the model context.

    Why it matters

    Invoicing and reconciliation are where most “AI assistants” quietly give up and route to a human. Giving your agent secure, scoped access to PayPal and your bank closes that gap — it raises the invoice, watches for the money, and answers from your real accounts, turning a polite chatbot into something that finishes the job and keeps your books honest while it does.

    Give your agent secure access to invoicing and banking →

  • Gmail, Drive & Calendar, handled by your agent

    Gmail, Drive & Calendar, handled by your agent

    Most of your team’s real work lives in Gmail, Drive and Calendar. So that’s exactly where your agent should be able to reach.

    An inbox your agent can actually use

    With the Google Workspace connectors, your agent can read and send Gmail — not just draft replies for you to copy, but handle the back-and-forth directly. It can find the thread a customer is asking about, pull the context, and respond, all without you switching tabs. The email stops being a wall between the conversation and the work.

    Files in, files out

    Your agent can read files from Drive and upload new ones to it, with real binary file I/O — so it handles actual documents, not just text. Need the contract a teammate referenced? The agent fetches it. Got a file in a conversation that belongs in Drive? The agent puts it there. The same handoffs that normally need a person are something the agent just does.

    Drive tokens are encrypted and stored in a vault. Your agent gets the access it needs to act, while the credentials stay locked away from the prompt and out of reach.

    A calendar that keeps itself

    The Calendar connector lets your agent manage your schedule as part of doing its job — so booking, checking and adjusting events isn’t a separate manual step bolted onto everything else. When a conversation produces a meeting, the calendar reflects it.

    • Gmail — read and send mail directly inside a conversation.
    • Drive — read and upload real files, binary and all.
    • Calendar — manage events without leaving the flow.

    Why it matters

    The difference between a chatbot and a teammate is whether it can reach the tools the rest of your team uses every day. Wiring your agent into Google Workspace means it can fetch the document someone needs or attach a file from a conversation — the small, constant tasks that quietly eat your hours.

    Connect your agent to Google Workspace →

  • Agents that generate images

    Agents that generate images

    Ask for a picture, get a picture. The hero images on this very blog were generated through Agentura with Grok — the same image plumbing your agent uses to make pictures on request.

    One connector, many engines

    The image-generation connector wraps several image APIs behind a single skill: OpenAI, xAI Grok, Google, Stability, fal, Fireworks and Together. Add an API key for the engines you want to use, and your agent generates images on request — no juggling seven SDKs, no seven separate dashboards inside the app. You pick the engine; the connector handles the rest. Leave the model unset and it uses that provider’s default, or name a specific one — pick Fireworks or Together when cost matters, OpenAI or Stability when you want maximum quality.

    From prompt to picture, in the chat

    When your agent makes an image, the result is saved to its workspace and returned as a signed URL — valid for about a week and ready to use — that can be shown inline right in the conversation. So a request for a social graphic, a product mockup or a blog header doesn’t end with a link buried somewhere; the picture appears where you’re already talking, in one call, no second round-trip to fetch it.

    We use it ourselves: the artwork at the top of these posts came out of Agentura, generated through Grok. If it’s good enough for our own blog, it’s ready for yours.

    Keys stay locked

    Pulling together this many image providers usually means scattering API keys across configs and hoping none leak. Here you add each provider’s key once in the Portal — it’s stored encrypted, and your agent calls the providers through the connector without ever holding the credentials in plain sight. Key up the engines you need and the secrets are handled for you.

    • Seven engines, one connector — add a key per engine you want; no per-vendor SDKs.
    • Saved & signed — results land in the workspace with a signed URL.
    • Inline in chat — the image shows up where the conversation is.

    Why it matters

    Visual work is constant — listings, posts, drafts, mockups — and it’s exactly the kind of small task that stalls when you have to leave the conversation to make it. An agent that can generate images on the spot keeps the work moving.

    Let your agent generate images on request →

  • Agents that read your PDFs

    Agents that read your PDFs

    A contract, an invoice, a 40-page report — the answer your customer wants is almost always already in the PDF. Your agent just needs to be able to open it.

    Turning documents into answers

    The pdf-tools connector gives your agent two straightforward tools — pdf_extract_text and pdf_info — to pull the text and metadata out of a PDF. Contracts, invoices, reports: as long as the file carries a real text layer, the content stops being a locked file and becomes something the agent can actually read, quote and reason about. (Scanned, image-only pages need OCR, which these tools don’t do.)

    Grounded in the document

    Reading a PDF is far more useful when the agent can connect it to everything else it knows. When you upload a PDF to your agent’s knowledge base, its text is extracted, indexed and retrievable — so the agent answers from the document’s real content, not a guess and not a generic summary. And when it reads a file directly with pdf_extract_text, that text comes straight back into the agent’s working context to reason over. Ask what a clause says or what an invoice totals, and the reply comes from the page itself.

    The point isn’t that your agent can read a PDF. It’s that it can read your PDF and answer the specific question buried inside it — with the rest of your knowledge in context.

    The work this replaces

    Think about who currently opens these files. Someone digs up the contract to confirm a renewal date, scrolls an invoice to find a line item, or skims a report to answer one question. With pdf_extract_text and pdf_info in hand, that’s an agent task — fast, repeatable, and available to your customers the moment they ask.

    • pdf_extract_text — pull the text out of a PDF that has a text layer.
    • pdf_info — read the document’s metadata at a glance.
    • Plus the knowledge base — upload a PDF and its content is indexed, so answers stay grounded in the document.

    Why it matters

    So much business knowledge sits trapped in PDFs that nobody has time to re-read. Giving your agent the ability to extract and understand them turns a pile of static files into a source it can answer from, around the clock.

    Put your PDFs to work with an AI agent →

  • Server automation over SSH

    Server automation over SSH

    A DevOps assistant that can only describe the fix is a tab you read at 2am. One that can run the command is the reason you stay asleep.

    A real shell, not advice

    The ssh-server connector gives your agent an ssh_run tool — the ability to execute commands directly on your servers. That’s the line between an assistant that suggests what to type and one that actually does it: checking a service, tailing a log, restarting a process, running the playbook. Your agent stops narrating ops work and starts performing it.

    Built for ops automation

    This is where an agent earns its keep as a DevOps teammate. Routine checks, deploys, health probes and the repetitive plumbing of running infrastructure all become things your agent handles on request — or on a schedule. The kind of work that’s too small to page a human for but too important to skip is exactly what ssh_run is good at.

    ssh_run is powerful by design — which is precisely why you scope it carefully. Give an agent access to the servers it needs, and nothing more.

    Power with a short leash

    A tool that can run anything on your servers deserves real discipline. Credentials are encrypted, and the connector pairs with need-to-know access control so each agent reaches only what its job requires. Treat the scope as part of the setup, not an afterthought, and you get serious automation without handing the keys to your whole fleet.

    • ssh_run — execute commands on your servers from a conversation.
    • Encrypted credentials — secrets stay locked, never in the prompt.
    • Scoped tightly — need-to-know access keeps the blast radius small.

    Why it matters

    Most operational toil isn’t hard, it’s just constant and it lands at inconvenient hours. An agent with a carefully scoped ssh_run tool can carry that load — acting on your infrastructure instead of merely talking about it.

    Give your DevOps agent a real shell →

  • Add any tool — no redeploy

    Add any tool — no redeploy

    The fastest way to give an agent a new skill shouldn’t be a release. With Agentura connectors, it isn’t.

    A marketplace and a gateway

    Agentura connectors are two things at once: a marketplace of ready-made integrations, and an on-the-fly call gateway for reaching APIs directly. Together they mean a new API can become an agent skill almost immediately. You’re not waiting on an engineering cycle to teach your agent something new — the capability shows up when you add the connector.

    No release required

    The old way to extend an assistant was to write the integration, ship it, and deploy. Here, giving an agent a new capability needs no redeploy. Add the connector, and the agent can use it. That changes the rhythm of building your team: when a need appears, you meet it the same day instead of putting it in a backlog.

    New APIs become agent skills almost immediately — no release, no downtime, just a new thing your agent can do.

    Scoped, secret-safe, multi-account

    Moving fast doesn’t mean moving loosely. Connectors are project-scoped, so each agent only reaches what its project owns, and secrets are always encrypted rather than sitting in the prompt. They’re also multi-instance: you can connect several accounts of the same type — two Stripe accounts, three mailboxes — and point the right agent at the right one.

    • Marketplace + gateway — ready integrations, plus direct API calls.
    • No redeploy — new capabilities go live without a release.
    • Multi-instance & scoped — many accounts, each agent kept to its project.

    Why it matters

    An AI team is only as useful as the things it can actually touch. When adding a tool is instant, scoped and secure, your agents grow with your business instead of lagging a quarter behind it.

    Add a new tool to your agent today →