The Consul Democracy on AWS by FOSSonCloud pattern now ships Consul Democracy 2.5.1, up from 2.5.0.
A note on scope up front: 2.5.1 is a maintenance patch — bug fixes and dependency updates — and it doesn’t have its own separate public changelog beyond that. Rather than invent detail, this post covers what 2.5.0 introduced and 2.5.1 is now running on, since that’s what actually matters if you’re arriving from 2.4.x or earlier.
What changed
2.5.0 (April 2026) — AI features and admin tooling
- AI-powered features, including LLM-assisted translations for proposals and AI-generated image suggestions for citizen-submitted content.
- Admin CSV exports for proposals, making it much easier to pull proposal data into spreadsheets for reporting or analysis.
- Legislation proposals joined the moderation area, so moderators can review them through the same workflow as every other content type.
- Valencian SDG (Sustainable Development Goals) images and translations were added.
- Maintenance: an upgrade to Rails 7.2, plus a batch of dependency updates and security fixes.
2.5.1 — maintenance
Bug fixes and dependency updates on top of 2.5.0.
Why it matters
The AI features are the change that needs a decision from you rather than just an upgrade.
LLM-assisted proposal translation is a real accessibility win for a participation platform. Multilingual civic engagement has always been bottlenecked on translation capacity, and a municipality that couldn’t afford human translation for every citizen proposal simply had monolingual participation. Machine translation changes the economics of that. But it also means a citizen’s proposal can be presented to other citizens in words the author didn’t choose — and in a democratic process, that’s a substantive question, not a technical one.
AI-generated image suggestions raise a similar issue: images attached to citizen proposals shape how they’re perceived, and a generated image is a synthetic framing of someone else’s idea.
Both of these need a human-review policy before they go live. Not because the features are badly built, but because “who is accountable for the translated text or the suggested image” is a governance question your moderation team should answer on purpose.
The less contentious changes are straightforwardly good. Bringing legislation proposals into the shared moderation area removes a workflow split that moderators had to remember. And CSV export for proposals closes a gap that had people scraping the UI or asking an admin for a database query every time they needed a report.
The Rails 7.2 upgrade is the invisible one: it’s what keeps the platform on a supported framework with current security patches.
What you should do
- If you’re coming from 2.4.x or earlier, this is your first exposure to AI-assisted translation and moderation tooling. Review it with your moderation team before it goes live.
- Write down a human-review policy for AI output. At minimum: is machine translation labeled as such to readers? Does a human approve a suggested image before it attaches to a citizen’s proposal? Who is accountable when a translation misrepresents intent? Decide this before the first complaint, not after.
- Check your locale coverage if you run a multilingual site, including whether the added Valencian SDG translations apply to you.
- Point your reporting people at the new CSV export. If anyone on your team has been manually pulling proposal data, this replaces that job.
- Confirm your moderators know legislation proposals now appear in the standard moderation area — otherwise they’ll keep checking the old place.
- Treat the upgrade itself as routine. 2.5.1 is a maintenance patch; the Rails 7.2 work landed in 2.5.0 and fresh deployments from the pattern carry it.
Version details in this post come from Consul Democracy’s own release notes. Where 2.5.1’s changelog wasn’t independently verifiable beyond bug and dependency fixes, this post says so rather than filling the gap. If you hit anything upgrading the pattern, open an issue on GitHub.
— FOSSonCloud