Skip to content
JDKTechnologies
Business AI Analyst

Find the work worth giving to AI, before you buy or build.

Many AI initiatives fail before the technology step. The problem is often choosing the wrong work to automate. The Business AI Analyst maps how work actually gets done, scores every task for automation fit, and hands you a ranked backlog, including a list of what not to automate.

Try the lens

How we score a task for automation fit

Every task in your workflows gets scored on dimensions like these. High scores go to the backlog. Low scores go to the kill list and stay with your people. Move the sliders to see how the recommendation changes.

Simplified preview. The real assessment scores 12+ dimensions per task, based on workflow interviews and data-readiness review.

TASK: INVOICE INTAKE & MATCHING

High
OccasionalConstant
Clear
Judgment callClear rules
Partial
ScatteredStructured
Low
ReversibleIrreversible
70
Strong automation fitGoes to the ranked backlog with an agent job description.
What you get

Deliverables, not decks

01

Workflow interviews

Structured sessions with the people who do the work, not just their managers.

02

Process mapping

How work actually flows across people and systems, exceptions included.

03

Data-readiness assessment

Whether the data an agent needs exists, is accessible, and is trustworthy.

04

Automation-fit scoring

Every task rated on repetition, rules, volume, data, reviewability, and risk.

05

Ranked use-case backlog

A prioritized list of agent candidates with effort and payoff estimates.

06

Agent job descriptions

Role, tools, permissions, and success measures for each candidate agent.

07

Risk & compliance review

Where approvals, audit, and data-handling constraints apply.

08

The kill list

Weak use cases named and retired before they burn budget.

09

First-agent recommendation

The single best place to start, with the reasoning on the record.

10

90-day roadmap

From first pilot to a managed foundation, sequenced and costed.

The kill list

Half the job is knowing what not to automate

Every engagement produces a “do not automate” list: work that’s too ambiguous, too high-stakes, too relationship-driven, or too exception-heavy for AI to own. That list protects your budget, your customers, and your team’s trust.

STAYS HUMAN-LED: EXAMPLES
Negotiating renewal terms with a long-standing client
Deciding a borderline benefits exception case
Delivering difficult feedback or personnel decisions
Approving spend that crosses policy thresholds
How it runs

From interviews to a 90-day roadmap

PHASE 1

Discovery sprint

Interviews, process mapping, and data-readiness review across target workflows.

PHASE 2

Scoring & backlog

Automation-fit scoring, ranked backlog, kill list, and agent job descriptions.

PHASE 3

Readout

First-agent recommendation and a 90-day roadmap your leadership can act on.

PHASE 4

Prove it

Optional: first agent pilot in a real workflow, measured against a baseline.

One Sprint. One ranked backlog. Zero guessing.

Start an AI Work Discovery Sprint