# AI use cases for Ciele (MattiaIppoliti/ciele) Repo: [MattiaIppoliti/ciele](https://github.com/MattiaIppoliti/ciele) | Use cases: 4 | Page: https://dedupedai.com/use-cases/mattiaippoliti-ciele/ Ciele is a self-hostable TypeScript platform for building embeddable AI chat assistants that answer from your own websites, files, and FAQs, with escalation to human help desks and conversation analytics. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Run a private AI assistant on your own machine: https://dedupedai.com/use-cases/mattiaippoliti-ciele/self-host-private-assistant-in-one-command/ - Build a support assistant from your website and FAQs: https://dedupedai.com/use-cases/mattiaippoliti-ciele/build-website-faq-assistant-with-escalation/ - Audit assistant answers from conversation transcripts: https://dedupedai.com/use-cases/mattiaippoliti-ciele/review-assistant-answer-quality-from-transcripts/ - Hand the Ciele install to a coding agent: https://dedupedai.com/use-cases/mattiaippoliti-ciele/delegate-ciele-install-to-coding-agent/ ## How to run a private AI assistant on your own machine with Ciele Repo: [MattiaIppoliti/ciele](https://github.com/MattiaIppoliti/ciele) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README Start the full Ciele stack locally with one bootstrap script and sign in as the owner. **Who it is for:** A developer or ops engineer who needs a self-hosted assistant platform and does not want to send content to a hosted Ciele service. **The problem:** Setting up a multi-part AI app with a database, migrations, and secrets by hand is slow and error-prone. **The outcome:** A running Ciele instance where the first account you create owns its organization. ### Steps 1. Clone the Ciele repository from GitHub (the README clone line is truncated here, so use the repo URL shown on the GitHub page), then run `cd ciele`. 2. Run `./deploy/bootstrap.sh` to generate secrets, start the database, apply migrations, and boot the app. 3. Open the app address shown in the output, sign up, and keep the account you create as the organization owner. 4. Optional: run `./deploy/bootstrap.sh --seed` to start from populated demo content. (inferred, check the README) 5. Point the app at a local server that speaks the OpenAI chat and embeddings API, using the settings described in the self-hosting docs at docs.ciele.app. ### Prompt to paste into your agent ```text In the ciele repository, run ./deploy/bootstrap.sh on , then report the URL where the app is running and whether the database migrations finished without errors. Verify that the app responds at that URL before reporting success. ``` ### Caveat The README's clone command is truncated in the source, so the exact repository URL must be taken from GitHub. Risk, credentials: The bootstrap script generates secrets for the database and app, so store the generated values somewhere safe and do not commit them to version control. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/mattiaippoliti-ciele/self-host-private-assistant-in-one-command/ ## How to build a support assistant from your website and FAQs with Ciele Repo: [MattiaIppoliti/ciele](https://github.com/MattiaIppoliti/ciele) | Difficulty: Intermediate | Time: 60 minutes | Basis: From the README Crawl your site, add files and FAQ pairs, and set a help desk for hand-off. **Who it is for:** A support lead whose customers ask the same questions and whose team needs a chat assistant that escalates when the answer is not in the knowledge base. **The problem:** Visitors ask repeat questions that answers on the website already cover, and the team has no controlled way to route the rest to people. **The outcome:** An embeddable chat widget that answers from your sources with citations and hands off to a help desk when it cannot answer. ### Steps 1. Run `./deploy/bootstrap.sh` and sign in to the app. 2. Create an assistant in the Assistants editor and set its welcome message, starter buttons, and answering style. 3. Add knowledge by crawling your website, uploading files, and curating FAQ pairs. 4. Create a help desk with its channels, forms, and availability windows so unanswered questions can be escalated. 5. Copy the embed snippet from the assistant and place it on your site, then test a question that is not in the knowledge base to confirm it escalates. ### Prompt to paste into your agent ```text In the ciele repository, set up an assistant that answers from and escalates to a help desk when no answer is found. Report the embed snippet and the result of one test question that should escalate. Verify that the escalated question appears as a handed-off conversation before reporting success. ``` ### Caveat Answer quality depends on the sources you add, and the README does not describe how crawl depth or file types are limited, so check results before publishing the embed. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/mattiaippoliti-ciele/build-website-faq-assistant-with-escalation/ ## How to audit assistant answers from conversation transcripts with Ciele Repo: [MattiaIppoliti/ciele](https://github.com/MattiaIppoliti/ciele) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README Read real conversations with citations and log answer-quality fixes in the tracker. **Who it is for:** A product or support operations manager who needs to know where the assistant gives weak answers after it has gone live. **The problem:** Without reviewing real chats, weak or missing answers stay hidden and the knowledge base never improves. **The outcome:** A list of conversations with weak answers, each linked to its sources and recorded in the improvements tracker. ### Steps 1. Open the conversation inbox and filter to conversations where the assistant escalated or answered poorly. 2. Read each transcript and check the citations to see which concept and source each answer used. 3. Record each weak answer as an item in the improvements tracker for answer quality. 4. Review the analytics view and insights for patterns, such as repeated unanswered topics. 5. Add or correct FAQ pairs and source files for the top patterns, then recheck the same questions. ### Prompt to paste into your agent ```text In the ciele repository, review the conversation inbox for and list the five weakest answers with their citations and the source each one used. Report them as a table with suggested knowledge fixes. Verify that each listed answer links to a real transcript before reporting. ``` ### Caveat The README says the tracker and insights exist but does not describe their filters or export options, so the review steps rely on what the UI shows. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/mattiaippoliti-ciele/review-assistant-answer-quality-from-transcripts/ ## How to hand the Ciele install to a coding agent with Ciele Repo: [MattiaIppoliti/ciele](https://github.com/MattiaIppoliti/ciele) | Difficulty: Advanced | Time: 60 minutes | Basis: From the README Let an agent clone, bootstrap, and verify a Ciele instance and report back. **Who it is for:** A team lead who wants an AI coding agent to set up a Ciele instance on a server or workstation without manual terminal work. **The problem:** Installation involves several steps and checks, and staff time is lost when someone has to watch each one. **The outcome:** A verified Ciele instance and a short report of the URL, the owner account step, and any errors. ### Steps 1. Give the agent the repository URL from GitHub and a target machine. 2. Have the agent run `cd ciele` and then `./deploy/bootstrap.sh`. 3. Have the agent run `./deploy/bootstrap.sh --images vX.Y.Z` instead if you want published images for a specific release. 4. Have the agent confirm the app responds at the address printed by the script and report any failed step. 5. Keep the owner sign-up for a person to complete, since the first account owns the organization. ### Prompt to paste into your agent ```text Clone the ciele repository from onto , run ./deploy/bootstrap.sh, and report the app URL and any failed step. Verify that the app returns a response at that URL before reporting success. ``` ### Caveat The README does not list the exact versions or tags to pass to --images, so the agent must take vX.Y.Z from the repository's releases. Risk, remote script: The agent will run a shell script from the repository, so review ./deploy/bootstrap.sh before letting an agent execute it on a shared or production machine. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/mattiaippoliti-ciele/delegate-ciele-install-to-coding-agent/