murmur | works — one form
For organisations that must understand what is happening in the public sphere — and want rhythm, evidence and action out of it.
UrbanStakes is one of several Murmur forms — the most mature platform, focused on urban dynamics & stakeholder relations. The form differs per challenge; the intelligence beneath — architecture, methodology, data sovereignty and governance — is the same.
Domain-agnostic UrbanStakes is suitable for all relevant societal policy areas, from discrimination & inclusion to social safety, energy transition, labour market and the Environment Act.
section b
An average policymaker is confronted with 800 messages per week. Media, motions, reports, complaints, social posts, letters. One person cannot possibly read that, let alone interpret it.
At the same time the board expects you to be on top of it. To know what is happening before the newspaper writes about it. To substantiate signals instead of acting on intuition.
The solution is not more dashboards or yet another tool. The solution is a swarm that listens for you.
section c
In one sentence, in three movements — listen, interpret, propose.
UrbanStakes scans hundreds of sources in your domain every day, links signals to the themes you track, and proposes what you could do about them.
01 · Listen
Hundreds of sources, curated per theme. Public and internal data in one fabric. Typical sources: municipal gazettes, motions, national media, FOI requests, research reports, complaint files (per L2/L3 rules), social channels. Adding new sources is an onboarding step, not a feature wish.
02 · Interpret
Signals are linked to themes, stakeholders, policy lines. Taxonomy is domain-specific, co-built with your organisation, not imposed. Each signal shows source, date, actors, relevant files, confidence score. A motion on social safety is linked to your casework of the past six months, to the alderman, to comparable motions elsewhere.
03 · Propose
Every morning a short brief with what matters and two or three possible responses. Proposals are specific: respond with letter X, call contact Y, put it on agenda Z. People decide. The platform illuminates, never executes on its own.
section e
National
National anti-discrimination infrastructure. Used for source curation, casework and thematic signalling.
In preparation
Q3 2026. We are looking for one or two municipalities willing to co-architect.
In collaboration with
section f
Four characteristics relevant to decision-makers. Technical deep-dive: see the Trust Center.
01
The platform listens 24/7. Failure of any single source or AI provider is automatically absorbed.
02
Not tied to a single LLM vendor. Models are interchangeable; the orchestration platform is not.
03
L0 public, L1 internal, L2 sensitive, L3 special-category. Each class determines which model may run where — see Security Model Selection.
04
Every AI call is recorded: which document, which model, which prompt, which response, which classification.
section g
Pricing follows a contribution-to-commons model — we discuss this in an exploratory conversation. We do not work with SaaS-per-seat; differentiation scales with contribution and domain complexity.
Annual contract, three months' notice, no lock-in.
section h
We walk through your themes with you. No scripted demo — we work live with your sources and your questions. If it doesn't feel relevant after 30 minutes, that's fine.
Book a demo