AI Assistant
A tool that processes information and generates suggestions based on patterns in data. It helps with tasks like summarizing, drafting, and organizing information—but does not make decisions or take actions on its own.
RIGHT HAND SERVICES × AI
Resources: Learning About AI-Assisted Workflows
Explore introductory material to help understand AI assistants versus AI agents, how to identify repeatable workflows, what to prepare before discussing automation, and why human approval matters.
AI recommends. JP decides.
AI ASSISTANTS VS. AI AGENTS
Understanding the difference
AI Agent (in the RHS context)
Combines an AI assistant with connected tools and a defined workflow. It can propose actions (e.g., “draft a follow‑up email”), but any action that modifies data or sends a message requires human review and explicit authorization before execution.
HOW TO IDENTIFY A REPEATABLE BUSINESS WORKFLOW
Looking for patterns in recurring work
Look for repetition
Do you perform the same sequence of steps for similar tasks? For example, every new lead follows the same intake, qualification, and follow‑up process.
Identify the trigger
What starts the workflow? It could be an incoming email, a form submission, a scheduled date, or a change in status.
List the inputs
What information or materials are needed at the start? This could be contact details, documents, or data from another system.
Map the processing steps
What happens to the inputs? Consider validation, transformation, summarization, or routing to another step.
Spot review points
Where does a person need to check the work before it proceeds? These are critical for quality and accountability.
Note authorized actions
Which steps actually change something in the outside world—like creating a record, sending a message, or updating a database? These require explicit approval before execution.
Record the outcome
What is left behind after the workflow runs? This could be a updated record, a sent message, or a completed task log.
WHAT TO PREPARE BEFORE DISCUSSING AUTOMATION
Making the conversation productive
Describe the problem, not the solution
Focus on what is frustrating, time‑consuming, or error‑prone about the current way of working. Avoid jumping to a specific tool or technology.
List the steps you currently take
Write down the workflow as you actually perform it, including any manual workarounds or shortcuts you’ve developed.
Gather example inputs and outputs
Collect a few real‑world samples of what goes into the workflow and what comes out of it—redacting any sensitive information.
Identify pain points
Note where the workflow slows down, requires rework, or depends on a single person’s memory.
Consider who needs to be involved
Think about who provides the inputs, who reviews the work, and who authorizes any actions that have external effects.
UNDERSTANDING A WEBSITE REVIEW
What to expect from a digital intelligence look‑at
It is not an audit or guarantee
A website review is a structured look at clarity, usability, and publicly available information—without promising rankings, traffic, or inclusion in AI‑generated answers.
It focuses on observable elements
The review considers what a visitor can see and interact with on the publicly accessible parts of the site.
It documents sources and uncertainties
Findings are backed by what is actually on the page or in public records, and any gaps or ambiguities are noted explicitly.
It produces a prioritized action plan
The output includes observations, supporting evidence, why it matters, suggested improvements, and a priority ranking—so you can decide what to tackle first.
READING A PROPERTY RESEARCH BRIEF
How to interpret the organized information
Distinguish facts from interpretations
A fact is something directly reported by a source (e.g., “the deed shows a sale date of January 15, 2020”). An interpretation is a conclusion drawn from that fact (e.g., “the property was likely vacant at that time”).
Note what requires confirmation
Items that need checking with a relevant authority or qualified professional are clearly labeled—for example, “Zoning needs verification with the planning department.”
Understand that gaps are normal
If a piece of information is not found in public records, the brief will note it as an unresolved question—not assume it does not exist.
Use the brief as a starting point
The organized information and question list are meant to save time for the professional who will conduct the deeper review.
WHY HUMAN APPROVAL MATTERS
Keeping accountability in the loop
AI does not decide
The AI assistant generates suggestions; it does not choose which suggestions to act on or whether to act at all.
Actions have effects
When a workflow step modifies data, sends a message, or interacts with an external system, it can affect real‑world outcomes. Human approval ensures that a person is accountable for those effects.
Review catches mistakes
Even the best‑designed assistant can misinterpret inputs or generate irrelevant suggestions. Human review is the safety net that catches those errors before they cause harm.
GLOSSARY OF COMMON TERMS
Definitions for useful reference
AI assistant
A tool that processes information and generates suggestions based on patterns in data. It does not make decisions or take actions on its own.
AI agent
In the RHS context, a combination of an AI assistant, connected tools, and a defined workflow. It can propose actions, but any action that modifies data or sends a message requires human review and explicit authorization.
Workflow
A sequence of steps that moves information from a trigger to a recorded outcome, with defined inputs, processing, review points, and authorized actions.
Trigger
The event that starts the workflow—for example, an incoming email, a form submission, or a scheduled date.
Inputs
The information or materials needed at the start of the workflow.
Processing
What happens to the inputs—such as validation, transformation, summarization, or routing to another step.
Review point
A checkpoint where a person evaluates the work before it proceeds to the next step.
Authorized action
A step that actually changes something in the outside world—like creating a record, sending a message, or updating a database. These require explicit approval before execution.
Record of outcome
What is left behind after the workflow runs—such as an updated record, a sent message, or a completed task log.
LET'S TALK
Have questions about AI‑assisted workflows?
Describe what you’d like to understand or explore. We can start with the problem, not a sales pitch.