Make local knowledge easier to find¶
At a glance. locali is more than a feed — it can become the searchable memory of your community. The optional intelligence features make that possible: Ask locali answers plain-language questions and backs every answer with sources, semantic search finds content by meaning rather than just keywords, the feed can rank by local relevance, and smart routing brings new content into the right spaces automatically. Everything is optional, designed for local processing (e.g. with Ollama), and set up by the operator under Governance → AI. In early versions these features are experimental.
Screenshot: an "Ask locali" answer with source citations — assets/screenshots/ask-locali-answer.png
What's behind this?¶
Everything that flows into locali — a council decision, an RSS article, a club event, a neighbour's help request — is more than a card in a feed. It is a piece of structured knowledge about a community.
locali can connect these pieces: surface topics, make relationships visible, and make information easier to reach through a plain-language search — even when it originally arrived as a municipal PDF or council minutes.
The quality of this depends on source availability, configuration and the model used. These features are optional and, in early versions, experimental — the hub stays fully usable without any AI.
In the default setup, this happens on your own infrastructure. Content or queries are only sent to external services for AI functions when an operator deliberately configures such providers.
For everyone / members¶
Ask locali¶
Ask a plain question — locali searches all imported content and shows you the answer along with where it comes from. In the app you find this feature as "Ask locali".
Example:
"When is Bahnhofstraße closing and for how long?"
locali responds:
The full closure of Bahnhofstraße begins on 12 May 2026 and is expected to last six weeks. The reason is the renewal of district heating pipes by the municipal utility.
Sources: Municipal bulletin no. 18/2026 · Council resolution 2026/0217 · Stadtwerke RSS announcement 28 Apr 2026
locali is designed to ground answers in sources from the hub and make those sources visible. Those sources can be both imported content (posts, events, municipal bulletins) and — in the hub-wide search — the profiles of organisations. When no sufficient content is found, locali is intended to make that clear. Answers should be grounded in the sources that were found and make that basis visible.
Each question stands on its own: locali deliberately keeps no conversation memory — there is no follow-up chain and no stored history. This keeps queries transparent and data-minimal.
This can simplify a lot: less keyword searching, less PDF-clicking, a faster path to the right sources. locali does not replace official information or the need to check the original source.
→ How to use Ask locali — step by step
Semantic search¶
If the operator has enabled semantic search, search understands meaning rather than just letters: a search for "childcare" can then also find content about nursery places or childminders without the exact word appearing in it. Semantic search exists hub-wide and — switchable separately — within individual spaces.
Relevant feed¶
The feed can rank content by local and topical relevance — without paid reach, advertising profiles, or commercial engagement logic.
A new council decision on cycle paths can therefore become more visible where space, topic, and source fit together — even if it has only just arrived. The source and the space always remain visible — it stays traceable why a post appears in the feed.
What makes this feed different from commercial platforms: the goal is relevance for you — not time-on-site, click-through rates, or advertising revenue for a platform. Content is not ranked lower because it generates little engagement. No content is promoted because someone paid for it. Your usage behaviour is not passed on to external services by locali.
Local memory¶
What arrives in locali stays findable — not just while it is visible in the feed.
A new member asks: "What has happened in the town centre over the last twelve months?" — locali can help with questions like this, because content is indexed and timestamped.
This is especially useful for:
- Newcomers who want to get their bearings
- Clubs that want to understand what has been decided in their area
- Engaged residents doing local research
For admins & operators¶
Two setup hubs, one master switch¶
The entire AI configuration lives under Governance → AI and is split into two hubs that can be set up independently and in any order:
- Semantic search & recommendations — everything around embeddings: semantic search, feed recommendations and semantic routing.
- Ask locali & writing help — everything around the language model (LLM): the assistant and writing aids.
Above both hubs sits a single AI master switch as the emergency stop: off means all AI features are disabled — the stored settings are kept. No toggle is ever "locked": you can always state your intent; locali shows the single most important open point next to it (e.g. "An embedding provider is missing").
Good to know: "Ask locali" needs both — language model and embeddings. The assistant answers questions from real sources; if the embeddings are missing, locali deliberately switches it to "not available" instead of inventing a sourceless answer.
Smart routing¶
Incoming content can be distributed not just by manually defined rules, but by what it is actually about. A municipal bulletin entry about a new playground on Mozartstraße? locali can identify such connections and support the assignment to relevant spaces or topics — depending on configuration, source availability and data quality.
The routing engine optimises exclusively on content relevance: which space is topically affected? Which content, tags, or areas belong together? It does not factor in popularity, ad relevance, or paid visibility — and operators can override it at any time with manual rules.
Knowledge gaps¶
Questions that "Ask locali" could not answer reliably are collected under Governance → AI → Knowledge gaps. This is not an error log but a map of missing content: it shows which sources you should add or which topics you should cover better.
Costs and responsibility¶
Cloud AI incurs API costs depending on provider and usage volume — with lots of content or frequent queries, it is worth checking the provider's billing. And as soon as external providers are used, content or questions can leave your own infrastructure: operators must review and take responsibility for provider, region, contracts and legal basis themselves. For data-sovereign operation, running Ollama locally remains the default recommendation.
How does locali index content?¶
locali reads all content that is imported into the hub — municipal bulletins, RSS feeds, posts, events. From this, structured signals can emerge about which topics appear frequently in a community and how pieces of content relate to each other.
The system indexes content — not personal user data. No profile is created, stored, or shared. Relevance comes from sources, spaces, tags, and semantic relationships — not from advertising profiles or commercial behaviour tracking.
Sources, limits and quality¶
Answers are grounded in the content available in the hub. Sources are shown so that users can check what an answer is based on.
When no sufficient sources are found, locali is intended to make that clear. AI can make mistakes. It does not replace official information, and it does not replace legal, medical, financial or governmental advice. Quality depends on source availability, configuration and the model used.
Data sovereignty as a guiding principle¶
The intelligence features can run on your own infrastructure. By default, local processing is intended — for example via Ollama. External cloud providers must be configured deliberately.
Operators who want to use cloud providers configure this explicitly in the AI settings. Questions are not linked to an account by default; optional aggregated usage statistics are a separate concern.
What locali is — and what it is not:
| This is locali | This is not locali |
|---|---|
| A retrieval system grounded in real community content | A chatbot that freely generates answers |
| A relevance signal for better content distribution | A commercial feed algorithm optimising for engagement |
| A search and Q&A tool for community members | A surveillance system tracking individual behaviour |
| A bridge between related pieces of content | A system that autonomously creates content |
| An optional helper layer | An AI platform or AI authority |
Frequently asked questions¶
| Question | Answer |
|---|---|
| Do I need AI to use locali? | No. All intelligence features are optional — the hub, keyword tags and routing rules work fully without AI. |
| Why don't I see "Ask locali"? | The operator of your hub has not (yet) enabled or set up the feature. |
| Does locali remember my questions? | There is no conversation memory and no stored history; questions are not linked to your account by default. |
| Do my content or questions leave the hub? | Only if the operator has deliberately configured a cloud provider. By default, local processing is intended (e.g. Ollama). |
| Are the answers reliable? | AI can make mistakes. Answers show their sources so you can check them — locali does not replace official information. |
| What happens when locali finds no answer? | locali is meant to say so transparently instead of guessing. For operators the gap lands under "Knowledge gaps" — a hint about which content is missing. |
Phased rollout
The intelligence features are introduced in phases. Which capabilities are available on your hub depends on the version deployed and the operator's AI configuration. → AI settings