EN
Idioma · Language · Lingua
How it works and how it is governed.
The right model for each task.
AbbiAI decides which model, which source and which tool to use at each moment. It is neither a proprietary foundation model nor an additional offer: it is the engine behind the seven offers.
-
Design principle
Selection by task
First, which model is enough for the accuracy, latency and context the task requires? Then, which one allows the sovereignty level the case requires? The lightest model that is enough also consumes less energy and water.
-
Design principle
Vendor independence
The architecture separates the entity, the profiles and the deployments from the model vendor and from the connected systems.
-
Design principle
Authorized sources
Each connected system is declared as an authorized source for specific data types, with a validity period.
Design principles and requirements. The delivered capability is specified in each implementation.
Seven consoles. Through configuration, not code.
A business expert governs Abbie and Caddie without touching the model. Configuration cascades down: no lower level can relax what the level above sets.
-
Console
Entity
Who the AIE agent is for each client: identity, personality, voice, global rules and regulatory framework.
-
Console
Training
What it knows and what it can do: sources, intents, procedures, permitted actions and test suite.
-
Console
Deployment
What it has to achieve, with whom, where and when: objective, audience, script, channels, scheduling and compliance.
-
Console
Evaluation
What it achieved and what gets corrected: dashboard, AI review, human scoring, alerts and approved learnings.
-
Console
«Actions»
Which operations exist and with what maximum limits. Only Abbiamo.ai publishes them.
See the seven consoles → -
Console
«Tasks»
Abbi Tasks combine published «Actions» into a sequence with a verifiable outcome; they are composed by Abbiamo.ai, partners and their clients.
Meet Abbi Tasks → -
Console
«Flow»
Chains «Tasks» into a complete process, with conditions, detours and the point where a person steps in. A «Flow» never expands what its «Tasks» and «Actions» allow.
Meet Abbi Flow →
The model writes. The rules decide.
When a language model doesn't know something, it fills the gap with something that sounds right: this is what the market calls hallucinations. In a bank or a hospital that is not a matter of style: it is a compliance failure. That is why, before answering, Abbie checks the sources in a fixed order and stops at the first one that has the data. A price, a balance, a rate, a deadline or a policy never comes from the open internet or from the model's memory. Every answer is recorded with the source that backed it.
-
Level 1
Your databases
What your data says right now, including the deployment's data: its customer base, its offers and its script. Example: a card balance or the status of a policy, as they stand today.
-
Level 2
Your knowledge base
Manuals, policies and rates your company approved, with a validity period. Example: this quarter's fee table. If the document has expired, it is not used.
-
Level 3
Your connected applications
Apps, microapps and management software your company operates with. Example: the available slots in the calendar or the reference number of a request in progress.
-
Level 4
Authorized official websites
Only by exception and when at least three agree. Never for a price, a rate or a balance. Example: the requirements for a government procedure or the financial authority's calendar of non-business days.
-
Level 5
The route you configure
If there is no backing, it says so and follows the configured route: a human agent or a rerouting process. Example: a customer asks for a rate that is not in any approved source. Abbie does not estimate it: she passes it to an advisor and tells the customer when they will get an answer.
See the full hierarchy in Abbi Studio →
The substance of every answer is deterministic: which sources, in what order, which data and when to stay silent. The wording may vary.
The emotional and the ethical are set in the entity. The cognitive is fine-tuned in each task.
-
1
Cognitive
Understands, remembers and diagnoses, anchored in approved data.
-
2
Metacognitive
Knows when it doesn't know; says so or escalates.
-
3
Linguistic
Converses clearly, handles objections and steers toward the objective.
-
4
Creative
Adapts the form and combines approved options; never makes up the substance.
-
5
Emotional
Functional empathy and self-control; never fakes emotions or uses them to pressure.
-
6
Moral and ethical
Explicit, auditable rules that no one relaxes.
-
7
Autonomous
Decides the next step and executes —or prepares— with Abbi Actions.
Talks on any channel. Acts in your systems.
An AIE agent is not a system: it works with yours. An intermediate layer with unique names: behind it, the work is done with what the client has, with what Abbiamo.ai provides or with a mix. Changing vendors does not touch the entity or the deployments.
-
Actions
Client system
When the client has the system and exposes an interface; for example, checking a balance in its collections system.
-
Actions
Catalog microapp
When the client does not have the system or does not expose it; for example, generating and sending a payment link.
Custom microapps → -
Actions
Mixed
One part lives with the client and another in the microapp; for example, recording the agreement and generating its document.
-
Contact channels
The ones you already use
Voice, WhatsApp, web chat, email, SMS and social media, depending on what each implementation enables.
-
Connectors
Protected credentials
Each connected system is registered with encrypted credentials that are never displayed.
-
Language
Connected systems
We talk about the function of each system —customer history, collections, telephony—, not brands. Software types (ERP, CRM, core) are named only to explain the difference between an AIE agent and a system, never as a requirement to buy.
AI that is accountable is designed, protected and demonstrated.
Security, standards and data sovereignty each have their own page under Solutions.
-
AI strategy
The why comes first
The belief, the six hows with their proof, what we sell and how to start with a first flow.
See the strategy → -
Security
Bounded, traceable mandate
«Actions» with limits no one expands, conversation locks, data isolated per client and change control.
See security → -
Regulation and compliance
Built for those who are held accountable
The standards we follow from the design stage, those of each industry and the decisions that always belong to your institution.
See the standards → -
Data sovereignty
Your data, where your case requires
Four deployment levels, from the cloud to your premises, and five questions to choose.
See the levels →
Let's see how it works in your operations.
Tell us which systems you have and what your regulator requires.
Let's talk about your operations →or write to us at hola@abbiamo.ai