Practices/05 AI agents

Answer every call.Book every lead.

Voice and intake agents that run in production: they answer the phone, book meetings, fill in your systems, and hand off to your team. Tested before launch and monitored after it.

Book a call
The front desk view: every call answered, what the agent booked, and what it handed to your team.

What we do.And how you will know it worked.

Every unanswered call is a customer calling someone else. Our agents answer, book, and hand off, day and night.

  • Since 1994 · Dallas, Texas
  • Salesforce certified consulting partner
  • ASTQB-accredited training provider
  • ISTQB CT-AI accredited

01

Voice agents

Today

After-hours calls go to voicemail, and most callers hang up.

With Rex Black Inc.

A voice agent that answers every call, books appointments, and sends your team a summary.

What you get

  • A voice agent on your number, trained on your services and policies
  • Bookings written straight into your calendar and CRM
  • A summary and transcript to your team after every call
  • A clean handoff to a person whenever the caller needs one

How you know it is done

Every call logged and reviewed, so you can see what the agent said and what it booked.

02

Intake and workflow

Today

Staff spend hours copying details from emails and forms into other systems.

With Rex Black Inc.

Agents that read every request, fill in your systems, and flag anything that needs a person.

What you get

  • Agents that read email, web forms, and documents
  • Fields filled into your CRM, ERP, or ticketing system
  • Exceptions flagged to a person, with the reason
  • An audit trail for every action the agent takes

How you know it is done

Accuracy measured against a sample your own team checks by hand.

03

AI guardrails

Today

AI pilots stall because nobody can say what happens when the model gets it wrong.

With Rex Black Inc.

Guardrails, monitoring, and tests for every agent, so you know how it behaves before customers see it.

What you get

  • Written rules for what each agent may and may not do
  • An evaluation suite run before every change
  • Monitoring and alerts in production
  • Governance aligned to the NIST AI Risk Management Framework

How you know it is done

A release scorecard for every agent change, held to the same bar as our software testing.

Measured results

All case studies →

0

Missed Calls

Professional Services

24/7 AI Phone Coverage

Zero missed calls with AI-powered appointment scheduling and FAQ handling deployed in 3 weeks.

Read the case study →

24/7

AI Availability

AI-Powered Intake

How it runs.Study, measure, build, own.

  1. 01

    Study

    Which calls, emails, and forms come in, what happens to them now, and which ones you lose.

  2. 02

    Measure

    Missed calls, response times, and hours of manual entry, so you know what the agent is worth.

  3. 03

    Build

    One workflow first, tested against real conversations before it goes live.

  4. 04

    Own

    Monitored in production, reviewed with you, and expanded once the first workflow is measured.

Questionswe hear first.

What happens when the agent does not know the answer?

It says so, takes the details, and hands off to a person. It is built not to guess.

Does it connect to our systems?

Yes. Calendars, CRMs, and ticketing tools, connected through their APIs and tested before launch.

How do you keep it on script?

Written rules for what it may do, an evaluation suite that runs before every change, and monitoring once it is live.

Which AI model does it run on, and what does that cost?

Each step runs on the least expensive model that passes an evaluation set built from your real cases. Hard steps keep a top-tier model; routine steps often run on models that cost a small fraction as much per task. When data has to stay under your own contract, open-weight models served by US providers or deployed in your cloud account are an option. Our AI model cost report in the resource library has the benchmark numbers.

Where do we start?

With one workflow, usually inbound calls or a single intake form. Once it is measured and working, we add the next one.

Free to read

Open the library →

Evaluating us? Start here · 9 reads

Putting an AI agent into production, what to read before the first call.

Four stops for leaders deciding whether and how to put an AI agent to work: when an agent is the right tool, how we choose models and test them before launch, what agents we run in production, and the questions to ask before you sign.

Start the path →

Learning path · 11 reads

AI and agents, from your first prompt to running agents in production.

Four levels, read in order: how AI models work and how to use them well, how teams choose between workflows and agents and pick the right model for each task, how production agents are built and tested, and how leaders sequence adoption.

Start the path →

Course · ISTQB CT-AI v1.0

The four-day course for testing AI-based systems, and using AI in testing.

Official outline for the ISTQB Certified Tester AI Testing course as delivered by Rex Black Inc. Four days. Eleven chapters. Covers testing AI-based systems and using AI to support software testing.

Course outline →

Talk · Workshop

Practical AI for small business.

A 90-minute hands-on workshop for non-technical small-business owners. Five real workflows you do every week (spreadsheet rescue, email drafting, voice-memo SOP, bulk FAQ writer, meeting actions), each with the exact prompt pattern, the right tool to use (Claude Artifacts, ChatGPT Data Analysis, Gemini in Google, Copilot in Office), and the two-minute verify habit that catches the common failure modes. Pick the workflow that solves a real problem on your plate this week, run the prompt, ship it before you leave.

Slide deck →

Guide

Seven questions to ask before you sign

Seven questions that show whether a delivery partner is set up to protect your launch. Ask every vendor, including us. Written for buyers who need to defend the choice to a CEO or a board.

Buyer guide →

Primer

How AI Language Models Work, in Plain English

What a large language model actually does: learning from text, predicting the next token, context windows, why models sometimes state false things with confidence, and why some models cost a hundred times more than others. No math required.

Article →

Interactive demo

Compare AI Models

Run the same prompt against up to 3 of 13 current models (Anthropic, plus open-weight models from OpenAI, Meta, DeepSeek, Mistral, Alibaba, MiniMax, and Z AI hosted in US data centers) side by side. See real cost, real latency, and real output quality, in real time. Built so you can stop guessing which model fits which job.

Interactive demo →

Talk · STRIVE Forum

Choose the right AI tool.

Part 2 of the STRIVE Forum AI miniseries. A 90-minute hands-on workshop for small business owners choosing between the dozen AI tools landing in their inbox every week. Two short concept blocks, two embedded labs, and a four-question rubric to grade any AI output before sending it.

Slide deck →

How many calls went to voicemail this week?Let's put a number on it.

Book a call