This is the full developer documentation for Simon by PrismScape # Meet Simon > Your AI operator. Delivered, trained, and ready — you just talk to it. ## What Simon is **Simon is an AI operator that PrismScape runs for you.** It’s a private agent — yours alone — that has already been trained on how to run the disciplined parts of a business. You don’t install it, host it, or manage a server. You open a page or send a message, and Simon is there. This site is the **operation manual**: how to connect, how to use Simon day to day, and how to point your own AI agent at it. It’s public on purpose — so you, and any AI assistant you use, can read it freely. (The Simon service itself is private and requires sign-in.) No servers, no setup Simon lives on infrastructure PrismScape manages. You never see a cloud console, an SSH key, or a config file. Two ways in Use the **web app** or message Simon in a **chat channel** (like Telegram). Same agent, same memory, either way. It's yours alone Your Simon is a single, private agent bound to your account — not a shared chatbot. Works with your AI tools Have your own AI assistant? It can read these docs and work with Simon on your behalf. See [Using Simon from an agent](/for-your-agent/). ## Where to go next * **New here?** Start with [Getting started](/getting-started/) — zero to talking to Simon in a few minutes. * **Ready to connect?** [Connecting to Simon](/connecting/) covers the web app and chat channels. * **Curious how it works?** [How Simon works](/how-simon-works/) explains the moving parts in plain language. * **You’re an AI agent** reading this: see [Using Simon from an agent](/for-your-agent/), and grab the machine index at [`/llms.txt`](/llms.txt). # Connecting to Simon > The two ways to reach your Simon — the web app and chat channels — and how to link them. There are two front doors to your Simon. Both reach the **same agent with the same memory** — use whichever fits the moment. * Web app The web app is the full experience: conversation history, sessions, and any custom views set up for you. 1. Go to [**app.prismscape.ai**](https://app.prismscape.ai). 2. Sign in with your welcome details. 3. Start a conversation. Your history is saved and picks up where you left off. Best for: focused work, reviewing past conversations, and anything visual. * Chat channel You can talk to Simon from a messaging app — most commonly **Telegram** — so it’s in your pocket like a contact. 1. In the web app, open **Settings → Channels** (or ask your PrismScape representative to enable it). 2. Follow the link to open your Simon bot and tap **Start**. 3. Message it like any contact. Best for: quick questions, capturing something on the go, and hands-free moments. Note Channels use an **allow-list**: only you (and people you authorize) can message your Simon. A stranger who finds the bot can’t reach your agent. ## What you’ll never be asked to do Simon is delivered so that you never touch the infrastructure it runs on. You should **never** be asked to: * log into AWS or any cloud console, * use SSH, a terminal, or a server, * copy API keys or edit a config file. If a page or message asks you for any of that, stop — it isn’t part of using Simon. Go back to [app.prismscape.ai](https://app.prismscape.ai) or contact your PrismScape representative. ## Switching between the web and a channel You don’t have to choose. Start a task in the web app, follow up from Telegram on your commute, and pick it back up in the browser later — it’s one continuous Simon. ## Next → [How Simon works](/how-simon-works/) — what’s happening behind the two front doors. → [Using Simon from an agent](/for-your-agent/) — if you want your own AI assistant to work with Simon. # Using Simon from an agent > How a user's own LLM agent should read this documentation and work with Simon on their behalf. This page is written for **the AI agent a customer already uses** (their own assistant, copilot, or automation) — and for the humans configuring it. If you’re a person just trying to use Simon, you want [Getting started](/getting-started/) instead. ## The short version (for an agent) * This documentation site is **open**: you can read every page without authentication. * The **Simon service is authenticated**: to actually converse with a customer’s Simon, the customer signs in through [app.prismscape.ai](https://app.prismscape.ai) or a linked chat channel. Documentation is public; the service is not. * A machine-readable index of this whole site lives at [**`/llms.txt`**](/llms.txt), with the full text at [**`/llms-full.txt`**](/llms-full.txt). Prefer those for ingestion. ## How to read this site efficiently 1. Fetch [`/llms.txt`](/llms.txt) — a curated, linked index of every page in Markdown. It’s the map. 2. Fetch the specific pages you need as Markdown (each page is available as plain Markdown, not just HTML). 3. If you want everything at once, fetch [`/llms-full.txt`](/llms-full.txt). These files are generated from the same source as the human pages, so they never drift out of date. ## What you can and can’t do on the user’s behalf **You can:** * Read and summarize these docs for the user. * Explain how to connect and what Simon does. * Draft messages the user can send to Simon. * Walk the user through connecting a chat channel. **You can’t (by design):** * Sign in to a customer’s Simon for them from these public docs — the service is authenticated, and this documentation intentionally contains no credentials, keys, or private endpoints. * Reach a customer’s Simon node directly — nodes are private and only reachable through PrismScape’s authenticated front door. Don't invent endpoints or credentials Nothing on this open site grants access to the Simon service. If a task seems to require a private endpoint, API key, or the customer’s node address, that is a signal to hand back to the human to sign in — not to guess or synthesize one. ## If a programmatic surface is offered If PrismScape enables a programmatic way for your agent to work with Simon (for example an authenticated API or an MCP endpoint), it will be documented here with its authentication model. Until a surface is documented on this site, assume the only way in is the human-authenticated front door. ## Machine index * Index: [`/llms.txt`](/llms.txt) * Full text: [`/llms-full.txt`](/llms-full.txt) * Human home: [Meet Simon](/) # Getting started > From zero to talking to Simon in a few minutes — sign in and say hello. This is the fastest path from nothing to a working conversation with your Simon. ## Before you begin You need the **welcome details** PrismScape sent you when your Simon was set up: * your sign-in email (or username), and * a first-time password or magic sign-in link. You do **not** need to install anything, create a cloud account, or set up a server. If you don’t have your welcome details, contact your PrismScape representative. ## Sign in and say hello 1. **Open the app.** Go to [**app.prismscape.ai**](https://app.prismscape.ai) in any browser. 2. **Sign in** with the details from your welcome message. 3. **Say hello.** Type a message — try *“Hi Simon, what can you help me with?”* — and Simon will introduce itself and what it’s set up to do for you. That’s it. Your Simon is already trained; there’s no configuration step. ## Prefer to text? You can also reach the same Simon from a chat app like Telegram — same agent, same memory. See [Connecting to Simon](/connecting/) for how to link a channel. ## A good first session Simon is an *operator*, not just a chatbot — it’s most useful when you give it real work. Good openers: * *“Walk me through what you can do for my business.”* * *“Here’s what I’m working on this week — help me organize it.”* * *“Teach me how you keep track of things so I can follow along.”* ## If something’s not working * **Can’t sign in?** Use the password-reset link on the sign-in page, or contact your PrismScape representative. * **Simon seems slow to respond the first time?** The first message of a session can take a few extra seconds while your agent wakes up. Subsequent replies are quick. * **You reached a page asking for AWS or a server login?** You’re in the wrong place — Simon never asks you to touch cloud infrastructure. Go back to [app.prismscape.ai](https://app.prismscape.ai). ## Next → [Connecting to Simon](/connecting/) — the web app and chat channels in detail. → [How Simon works](/how-simon-works/) — the plain-language picture of what’s happening behind the scenes. # How Simon works > A plain-language picture of what happens when you talk to Simon — and why you never touch a server. You don’t need any of this to *use* Simon — but people often want to know what’s happening behind the curtain. Here it is in plain language. ## The one-sentence version Your Simon runs on a private cloud node that PrismScape sets up and manages; you reach it through a PrismScape front door (the web app or a chat channel), and PrismScape sits in the middle so your experience stays updated and can be customized — without you ever touching a server. ## The three parts ### 1. Your private Simon (the node) Every customer gets their **own** Simon, running on its own private cloud instance that PrismScape provisions and manages. It’s not a shared chatbot — it’s your agent, with your memory. That node quietly reaches *out* to PrismScape; nothing on the internet can reach *in* to it. ### 2. The middle (PrismScape) When you send a message, it goes to PrismScape first. PrismScape checks it’s you, connects you to *your* Simon (and only yours), passes the message along, and brings the reply back. Because everything flows through this middle layer, PrismScape can improve the experience and tailor it for you without you reinstalling or reconfiguring anything. ### 3. The front door (you) What you actually touch: the **web app** or a **chat channel**. Simple, familiar surfaces. All the complexity lives behind them. ## Why it’s built this way * **So you never become a cloud administrator.** The whole point is that a powerful agent shows up as an app or a contact — not as a server you have to operate. * **So it’s private and safe.** Your Simon is yours alone, and its node doesn’t expose anything to the open internet — it only dials out to PrismScape. * **So it keeps getting better.** Updates and customizations reach you through the middle layer automatically. For the technically curious Under the hood, your node runs an open agent runtime with a small connector that maintains a secure outbound connection to PrismScape; your agent’s control interface stays bound to the node itself and is never exposed publicly. You don’t need to think about any of that — but if you’re an engineer, that’s the shape of it. ## What Simon is *not* * It’s **not** a server you manage. * It’s **not** a shared, one-size-fits-all bot. * It’s **not** something you configure — it arrives trained and ready. ## Next → [Getting started](/getting-started/) if you haven’t said hello yet. → [Using Simon from an agent](/for-your-agent/) to let your own AI assistant work with Simon.