Capabilities
Three lists, kept apart on purpose.
A feature grid is the easiest place on any product site to blur what a product guarantees with what it merely offers. So this page does not have one. It has the guarantees the architecture makes, the capabilities built on top of them, and what has been decided against permanently — and it never lets a reader mistake the third list for a gap a large enough contract could close.
List one
The guarantees underneath everything.
Every line here is an architectural fact or a permanent commitment: a field that does not exist, a key that was never created, a floor the product will not move. Not one of them is a per-customer setting, so not one of them is available to negotiate. How the separation is built, and what each domain holds.
- No account, anywhere in the anonymous experience
- There is nothing to sign in to, so nothing accumulates against a person. The anonymous experience never stores name, email, phone, IP address, device identifier, fingerprint, session identifier, cookie identifier, GPS or location, free text, diagnosis, risk score.
- Three data domains with no join key
- Anonymous activity, voluntary contact identity, and platform configuration are separate systems. No key exists to connect the first two, so no query and no future feature can follow one to the other. What a key cannot do, low volume still can — which is why the two floors sit between every record and every reader.
- A reporting floor of ten activity events
- Below ten activity events, a value is suppressed rather than shown — and the floor is applied to each topic in each hour of each day, before the month is added up. The floor is uniform across every topic and every deployment, and no per-customer threshold field exists to soften it.
- A population floor of fifty people on any reportable unit
- A placement or audience pack becomes its own row only when the population it covers is at least fifty people. Below that, its activity rolls up into its parent rather than narrowing toward individuals.
- No individual risk score, at any tier
- There is no field to hold one, no model that produces one, and no plan to add either. The same is true of session history and of any composite score about a human being.
- Choices instead of anonymous free text
- The anonymous side offers statements to select. Because there is no text box, there is no narrative to store, subpoena, misread, or quietly analyse later.
- English and Spanish, authored in parallel
- Neither language is a translation layer over the other. Every page of this site exists in both, and the Spanish is written for the people reading it rather than converted sentence by sentence from the English — which is why the two differ in shape wherever the languages do, and never differ in what they claim.
- Crisis resources that a customer cannot remove
- An organization can add its own points of help. The crisis resources HABLA ships, including 911 and 988, are always there and cannot be removed. Interacting with one produces no organizational record: no count, no trend, no alert, and no row in any report.
- Pricing that does not move with distress
- The annual price follows deployment — sites and audience packs — and does not increase because more activity, more topics, or more contact requests occur.
- No behavioural tracking on this site
- No analytics SDK, no session replay, no fingerprinting, no ad pixel, no SVG-interaction logging, and no third-party font or script request on any page of it.
- What is recorded instead
- The anonymous experience keeps records, and this page would be dishonest without saying which: the deployment, the placement, the audience pack, the topic, the date, the whole hour, the language, a count, and how much the statement fit — nine fields, all on one row. Two more records are kept beside that row rather than on it, both of them counters: how many topics an interaction involved, and how often a topic was named the most important, which is a count for a whole calendar date and carries no hour. Every one of them is about an interaction. None is about a person.
List two
What HABLA does with what it has.
These are the capabilities the guarantees above are built to carry. Each names the part of the product it belongs to, and each states its own limit in the same breath as the claim, because a capability described without its limit is exactly the thing this page exists to avoid.
The person’s side
Type-ahead topic search that stays on the device
HABLA lets someone type a few words to narrow a topic list. That search runs on the device, and what is typed never leaves it.
The person’s side
Self-described fit
HABLA asks how much a statement fits — a little, quite a bit, very much — never how severe something is. It is reported as a distribution and never averaged into a figure about a person.
The report
Topics per interaction
HABLA reports how many topics an interaction involved, as a distribution. It is a shape, not a depth, and it is never rendered as a severity or a rank.
The report
Priority, separately from frequency
HABLA shows which statement was chosen as most important, separately from which was chosen most often. The two answer different organizational questions and are never combined into one score.
The action record
A record of what the organization decided
A signal carries the organization’s own decision record, and these are all eleven of its fields: a leadership question, a decision, an owner, an action, the date the action starts, a completion state, the topic the decision is about, whether a standing watch is open on that topic, which strategy record the action was taken from, whether that strategy was generated rather than written by a person, and the leadership notes the organization keeps for itself — free text an administrator writes, which stays inside the console and is never sent to a model. Those eleven are the record’s content; the row carries two more things and both are administrative, the reporting period it belongs to and the dates HABLA needs in order to retire it. The owner is a named administrator from the organization’s own console, and the Action Center shows that person’s name; nobody from the anonymous side appears in the record at all. When that administrator is deactivated the record keeps the decision and drops the person — the name is removed, and it is replaced by the role only where more than one account holds that role, and otherwise the decision keeps only the organization. It never asserts that an action caused an outcome.
Sector and national data
Privacy-safe benchmarks
HABLA compares an organization against privacy-safe sector and national aggregates. Every benchmark row carries its denominator, because a bare percentage reads as a share of people and HABLA does not measure people — and a difference may reflect deployment age, awareness or adoption rather than anything about the people. A deployment without enough history is told so rather than shown a comparison it cannot support. No cross-organization figure has been published yet, and none will be until one clears all three contributor floors.
Sector and national data
A contributor floor on every cross-organization figure
A cross-organization figure is published only where at least ten organizations contribute, no single one exceeds 40%, and the top three do not dominate it. No such figure has been published yet, and none will be until one clears all three — the combined limit is a number HABLA has not set, and the code refuses to publish rather than assume one.
Signal intelligence
Aggregate outlook, with its uncertainty attached
HABLA forecasts aggregate signal movement — never human behaviour, diagnosis, crisis, or individual risk. Every outlook carries a confidence range, and a deployment without enough history is told it has insufficient history for an outlook rather than given a number.
Administration and content
Customer-authored topics and resources
Customers add and reword their own resources and search aliases every month, in every plan, and write topics of their own from Professional upward — an Essential deployment uses the topic library rather than authoring topics into it. The flexibility applies to what an organization asks about — never to what HABLA is able to know about a person, which does not change between plans and is not a setting.
Sector and national data
Licensed access to aggregated intelligence
What HABLA sells through its API is licensed access to aggregated intelligence — never the sale of people’s behavioural data. The architecture forecloses the alternative: no raw-event endpoint and no per-person endpoint exists to sell.
List three
Refused, and not on any roadmap.
These are not missing. They were considered and decided against, each for a stated reason. A buyer comparing products needs to know which absences are shape and which are permanent — and these will not arrive later for a large enough contract.
- Engagement rate as a measure of success
- It would make distress the thing the product is optimised to produce.
- Unique users
- HABLA cannot count people. A number that claimed to would be an invention.
- A wellbeing index, or any composite score about human beings
- A single number about people invites exactly the comparisons the product exists to prevent.
- A per-tap date and time in any report
- In a small unit, one timestamped event plus ordinary workplace knowledge identifies a person.
- Which topics were selected together
- Combinations narrow toward individuals faster than any single topic does, and no threshold has been designed that makes it safe.
- A ranking of organizations by risk, distress, or wellbeing
- HABLA will not build it. “Largest change in reportable activity” compares signals; “highest-risk schools” is a judgement about places and the people in them, and no threshold makes that safe.
- Real-time monitoring, alerting, or crisis oversight
- HABLA is not monitored in real time, and a product that implied otherwise would fail the person relying on it at the worst possible moment.
- Any lookup from a contact request back to anonymous activity
- There is no key, so there is no query. This is the boundary the whole architecture is arranged around.
The two ends of the list are the same promise.
What HABLA builds and what it refuses to build are decided by one rule: does this move HABLA toward knowing more about an individual person?