Blind decisions
Decisions made on experience or gut feeling, with no data behind them.
Judgment Audit · Judgment diagnostic
Your company is going to implement AI. The question is where, how and with what judgment.
In one week we audit how one area of your business makes decisions, codify the judgment of the people who decide and deliver a prioritized plan to implement AI where it actually moves results.
5 days
business days
1 area
of the business
7–9 h
of your team’s time
Where value gets lost
Four signs your decisions don’t run on a system.
Before choosing tools you need to know which processes gain from AI, which need data or order first, and which of your company’s rules AI has to respect.
Data that doesn’t decide
There are reports and dashboards, but they don’t change any decision.
Routine that drains
Repetitive decisions with clear rules that a person still makes by hand.
Judgment in one person
When that person is out, the process stops.
Our thesis · The Judgment Layer
AI executes.
Judgment decides.
Before implementing any tool, you need to understand and codify how your company decides. That layer of judgment is what turns AI into an advantage of your own.
How it works
Five moments. One condition to move forward in each.
The week starts when we have the complete material, not when the contract is signed. That’s how we protect the promise: a plan in five business days.
Before day 1
Prepare
Questionnaire, 40 to 60 real cases and read-only access.
To move on: Material complete
Days 1 and 2
Expose
Interviews and a map of every decision in the area.
To move on: Critical decisions identified
Day 3
Codify
Workshop with your expert: labeled cases, rules and exceptions.
To move on: The expert validates the judgment
Day 4
Prioritize
Initiatives scored by impact, feasibility and time to value.
To move on: Each initiative has a metric and an owner
Day 5
Decide
Session with whoever owns the budget.
To move on: Decision recorded
What you take away
Six deliverables that stay with your company, whether or not you keep working with us.
Decision Map
A map of every decision in the area, sorted into four types.
It helps you
See at a glance what’s decided blind, what depends on one person and what can be automated.
Golden Set
20 to 30 real cases labeled with your best expert’s judgment.
It helps you
Train and evaluate any AI against your standards, not generic answers.
Judgment document
The rules, exceptions and limits your expert decides with, in writing.
It helps you
Make sure that knowledge doesn’t depend on one person and can be taught or automated.
Comparative demo
AI solving your cases with and without your judgment, with its accuracy rate.
It helps you
Measure what codifying your judgment is worth before investing in implementation.
Judgment Blueprint
3 to 5 prioritized initiatives, each with a metric, owner, scope and risks.
It helps you
Know what to do first and justify the investment to leadership.
Next-step proposal
Three possible paths forward, with scope and timing.
It helps you
Decide how to move on with all the information on the table.
The proof
One case, three answers.
Should a product piece with the price shown large be approved for an awareness campaign?
AI without your judgment
Approve
It follows the brand guidelines: correct colors, logo and typography.
Wrong
AI with your judgment
Reject
It applies your expert’s rule: in awareness, price doesn’t lead the piece.
Correct
Your expert
Reject
The reference answer the AI is measured against.
Correct
In the audit this is repeated with 20 to 30 real cases and we measure the accuracy rate of each version.
Where to apply it
One method, four entry points.
Each area has its own questionnaire, cases and data sources. The method stays the same.
Sales
Which lead to prioritize, which discount to approve, when to let an opportunity go.
Marketing
Where to invest, which campaign to scale or cut, how to read attribution.
Operations
What to assign to whom, which exception to approve, when to escalate a problem.
Reputation
Which mention to answer, when to escalate to a crisis, what to say as a spokesperson.
Also applies to
Data and measurementStrategyCustomer serviceProjects
The method comes from projects we’ve already built.
Reputation
Longitudinal media monitoring
Tracks what the media publishes about an organization over time and classifies each mention with AI, using 14 rules that never break and 40 hand-labeled cases.
How it helps: The communications team stops reading everything and focuses on the mentions that call for a decision: respond, escalate or let it go.
Marketing
Brand value metric
Net Build Rate, a proprietary metric that measures whether a brand is building or losing value, developed with a global sportswear brand in Mexico.
How it helps: Turns brand value into a number you can track over time and defend to leadership, beyond reach and impressions.
Sales
Sales copilot on WhatsApp
Takes in the leads that arrive on WhatsApp, scores them with AI and suggests a priority and a reply to the salesperson.
How it helps: The team serves the highest-potential leads first, and a person always makes the final call before replying.
Data
DataOps platform with a voice layer
Connects business data to dashboards and to a voice layer for asking questions in natural language. Licensed to a creative group.
How it helps: Leadership checks its indicators whenever it needs them, without waiting for an analyst to build the report.
Teams
AI adoption at a creative agency
Role-based training with skill templates and a prompt library for content, creative, strategy and operations.
How it helps: Each team leaves with tools set up for its daily work, not general theory about AI.
Teams
AI training for mass transit
A 10-session, 20-hour program for the teams of a transit company.
How it helps: Designed so each team decides which tasks to delegate to AI and how to validate what it produces.
After the audit
You decide the next step.
Path A
Execute with A:BRA
We implement the prioritized initiatives. Content and social, with our partner MTM.
Path B
Execute with your team
We support you and transfer the judgment so your team can run it.
Path C
Wait
We review the plan again with your team after 30 days.
Why do it
One week to stop implementing AI blind.
You come out knowing where AI moves results, what to automate first and with what judgment, before committing budget to tools.
Next step
AI is now within everyone’s reach. Your judgment is what no one can copy.
Pick the area you want to start with and we’ll schedule the kickoff. One week later, you have the plan.