Dreamcatcher
Dreamcatcher is the AI-tool evaluation engine we built so we'd stop saying yes to platforms that don't serve our community. It works because it's tuned to BLKOUT's tech stack and our core values. Here's how it actually runs — and what's transferable as a pattern, even though the software isn't.
// the problem
There are dozens of AI tools, platforms, and services pitching themselves to community organisations every week. All of them come with a sales deck. Most of them won't survive contact with the questions we actually need to ask.
We were tired of the cycle — saying yes to something because the demo looked good, then realising six months later the data sovereignty was wrong, or it didn't deploy in our stack, or it served people who weren't our community. Tired of wasting time and money on decisions we'd reverse.
Dreamcatcher is the council we wished we had: a small set of named voices that read whatever's on offer against the same value base every time. The point isn't to make the decision for us — it's to keep us on course.
// the council
Every decision goes through a council of named voices — Baldwin, Murray, Rustin, Rivera, Newton. Each asks one specific question. The names aren't decorative. Naming the question after a person who held a particular line in real life makes the question harder to wave through.
Baldwin · the critic
"What's the price we don't see?" Lock-in, dependency, the cost when this tool changes its mind.
Murray · the ethicist
"Where does our data live, and who reads it?" Data sovereignty, training, surveillance.
Rustin · the builder
"Will this actually deploy in our stack?" Technical fit, maintenance load, what'll break.
Rivera · inclusion
"Who is this tool not for?" Accessibility, register, geographic reach, cost-to-attend.
Newton · the collaborator
"Where's the cooperation we're missing?" Reads across decisions — surfaces when other orgs are asking the same question and shared work is possible.
// example
The judges work from a URL — you submit the thing being pitched at you, and each one reads it in turn.
Input — Tool X · a commercial event-ticketing SaaS pitched at LGBTQ+ organisers in the UK. Should BLKOUT use it for community events?
each judge returns several hundred words of analysis. these are the summary verdicts — the full outputs live in the dreamcatcher app.
Baldwin · the critic
Third-party SaaS where BLKOUT becomes a customer, not infrastructure owner. Attendee data and payment processing live in their system. Duplicates the RSVP capability we've already built. Solves a problem we don't have while creating problems we're trying to avoid.
Verdict — Pass.
Murray · the ethicist
Commercial platform whose business model depends on data aggregation. Marketing copy emphasises charitable support to the community, but there's no evidence members can audit what's held about them, or that they control how it's used.
Verdict — Pass.
Rustin · the builder
Not self-hosted — nothing to deploy. Walled garden: no integration with our CRM, member pages, or AIvor. Member data collected through the platform stays in the platform. The actual constraint we have is programming and promotion, not ticketing mechanics.
Verdict — Pass.
Rivera · inclusion
App-first, smartphone-required, image-heavy — barriers for people in temporary housing, rural areas, or with data-cost constraints. Heavily city-centric. For closeted people or those in unsafe households, identity-flagged event titles in a browsing history are a real risk. Serves people already out, already networked, already in major cities.
Verdict — Pass.
What we decided — four-way pass. We didn't adopt it. The judges named what we were already half-thinking but hadn't articulated, and gave us the language to say why — to ourselves and to the partners asking.
// the pattern
The software doesn't travel — your stack and values aren't ours, and shouldn't be. The shape does:
// what it doesn't do
// the cost
// invitation
We're not productising Dreamcatcher. What we offer is conversation — about what AI-assisted ethical decision-making could look like inside your stack, against your values. Bring a decision you're stuck on; we'll bring what we've learned.
Say yes to a conversation →