A person can argue about homelessness on a large social platform for ten years. Nothing in that platform connects the argument to a person who acts on homelessness. Conversation is the product there, and attention is the revenue. engage.re connects a conversation to the action that follows it, and it records the connection. The same platform is the foundation under every system we build for a client.
engage.re is two things at once, and both are true of the same software.
It is a social platform where a conversation leads to a recorded action. It is also the platform on which we build client systems. Sense Future built it, and it has run in production since December 2025. Its sovereign contract has been live since 6 July 2026.
ESRE Media is the consultancy arm of Sense Future. We work in Enfield, London, and we build your system on this platform.
A conversation on engage.re can produce a commitment, and a commitment is a record.
A person raises a problem. Other people respond. One response becomes a commitment to act, and the platform holds that commitment as data linked to the conversation that produced it.
Two properties follow. You can trace an outcome back to the discussion that caused it. You can also measure which discussions produce outcomes and which do not.
Connect with us about your project. Message on WhatsApp Replies within 24 hours
engage.re has no likes. It has eight reactions, and each one carries a different weight.
Each reaction states what a contribution did to a conversation. The weights therefore measure contribution rather than volume.
A person who speaks once with the most useful input scores above a person who posts all day and adds nothing. A contribution that divides a discussion reduces a score, and no amount of activity raises it.
Three safety properties are structural on this platform, so no moderation policy is needed to hold them.
The third property is the one that gives the other two their value. A supervision record a supervisor can edit records a claim about the past.
Every action on the platform writes an event. Each event carries a hash of the event before it, and each entry is signed.
An alteration to one entry therefore breaks every entry after it, so the change is detectable. The log states what happened, rather than what somebody chose to leave in it.
One authorisation decision covers every crossing of a domain boundary. One place decides, so you can state the policy by reading one thing.
Most large systems hold their access decisions in hundreds of places. Each check made sense when a developer added it, and the set as a whole then holds no author. One production estate on this platform replaced 219 scattered checks with a single gate.
Permissions are capabilities rather than roles. Each grant names one action on one scope, and three rules apply to every grant.
Live production has issued 57,267 of these grants.
Identity is a key, and the platform stores no password. A person signs in by proving control of a key, so there is no password store to steal.
One model therefore covers every actor. Your staff, your applications and your AI agents all authenticate the same way, and each one is nameable in a log.
Each record is encrypted separately, under a key you hold.
Three consequences follow, and each one answers a duty under UK data protection law.
A breach of the store yields ciphertext, and a lawful order to us produces ciphertext.
Private messages are encrypted end to end. Sense Future cannot read them, and the platform holds no mechanism that would allow it.
The platform holds the meaning of every record type in one shared dictionary, and it holds that meaning as data.
An application declares its vocabulary before it writes anything. Every record type, every field and every relationship. The server then refuses any write outside that declaration.
Four properties follow.
Independent research measured what recorded meaning does for a language model. Three frontier models answered 99 business questions with the schema alone and scored 45.5% to 50.5%. With a short document describing the measures, all three reached 67.7% to 68.7%, and the three then became statistically indistinguishable from each other.
Five database tables hold everything, and those five never change.
A new record type adds an entry to the dictionary. No table appears, and no migration follows. Your new service line, your merged teams and your new funder requirement are therefore configuration rather than development.
Six interfaces carry everything, and they hold 61 operations and 39 query types. They work over JSON, from any language.
Two performance figures matter to a user. A read inside a domain returns under 200 milliseconds at the 95th percentile. A change reaches a reported figure within 60 seconds.
Four practices apply to every change, and each one produces evidence you can inspect.
The declared vocabulary does part of this work by itself. An expectation held as a declaration is machine-checkable, and the server enforces it on every write.
Three deployment tiers are available. We host it, you host it, or you run it on your own infrastructure.
The self-host route is real rather than nominal. Sense Future publishes a conformance package at engage.re/conformance, at version 2.0.0, which states the behaviour a conforming implementation must show. A third party can therefore build a replacement and test it.
Your exit is a design property rather than a promise. Your meaning is data, so an export carries the definitions with the records. You hold your keys, so your access does not depend on our cooperation. The specification is published, so a conforming implementation is possible without us.
Sense Future built engage.re to solve one problem. A conversation and the action that follows it must sit in the same record, under the same meaning, with the same accountability.
The properties that solved that problem turned out to solve a wider one. An organisation faces the same requirement as a conversation. It holds records in several systems, each with its own idea of a person, and it cannot trace an outcome to a cause.
Sense Future sets out the wider thinking in Machines of Human Grace.
Bespoke software normally carries three costs. It takes a long time. It becomes legacy. It leaves you dependent on the people who built it.
This platform removes each of those costs by construction.
Your system can therefore be as specific to your process as you need, without the usual consequences.
Your system arrives with these properties on its first day, and none of them is a feature we add later.
On top of that sits your own system. Your record types, workflows, screens and obligations, built in the shared vocabulary, at whatever weight your sector needs.
React 18 · TypeScript · Vite · PostgreSQL · Node.js · Express · WebSocket · End-to-end encryption · engage.re Graph API