What Real Chatbot and Voice Bot Usage Looks Like in Shipyard Staffing

If you run a staffing agency that places Ship Fitters, Structural Welders, Marine Electricians, or Shipyard Machinists, you have probably been pitched an AI chatbot or a voice bot at least a dozen times by now.

And you have probably ignored most of those pitches.

Fair. Most of what is out there is a glorified FAQ widget or a phone tree with a synthetic voice. That is not going to move a hiring surge for a Pascagoula shipyard or a winter fleet refit.

This post is different. It is not a pitch for AI in the abstract. It is a look at what real chatbot and voice bot usage looks like across two consecutive months of live campaigns in shipyard and aerospace staffing — the conversations, the numbers, and where they showed up in the actual hiring pipeline.

The workflow we will be talking about is TalentPrism’s 24/7 candidate concierge, Hannah. Hannah greets candidates on the website, answers inbound phone calls, runs structured intake, matches candidates to open reqs, and routes warm handoffs back to recruiters. Same workflow, three surfaces: web chat, phone line, and database reactivation outreach.

Why shipyard trades make this hard

Shipyard trades are some of the worst-fit candidates for a standard chatbot.

They are not sitting at a desk refreshing LinkedIn. They are on dry dock, under a hull, welding in a compartment, or on a fleet refit out of town. They answer on their own time, often after hours. They trust a recruiter they have worked with before and they tune out everyone else. A typical “Hi, how can I help you today?” chat widget does nothing for this audience.

So the bar for AI in this space is higher than “someone typed into a chat window.” The bar is:

• captures candidate intent when recruiters are off
• collects trade, pay, location, availability without recruiter back-and-forth
• routes real candidates to real open reqs
• shows up in submissions and hires, not just dashboard metrics

Here is what the last two months of data actually looked like against that bar.

Period 1: Jan 14 – Feb 13, 2026 (shipyard trades focus)

During this period, a national staffing firm ran four outreach campaigns through TalentPrism, three of them shipyard:

• Machinist
• Marine Electrician — Winter Fleet
• Structural Welder — Winter Fleet
• Ship Fitter — Winter Fleet

Across those campaigns:

• 21,854 outreach messages sent
• 75.3% blended deliverability
• 3.9% first-response rate
• 26.5% positive-intent rate among responders

On the conversational side, that outreach fed into both Hannah on the web and Hannah on the phone:

• 13 phone conversations handled by Hannah
• 191 web chat conversations
• ~5.75 hours of active web chat time
• 77% engagement rate on the largest web chat channel

This is not a sandbox demo. This is candidates texting back after a night shift, clicking the career site at 10pm, and calling a number to ask whether the welder role is first or second shift.

Period 2: Feb 14 – Mar 13, 2026 (aerospace focus, but the engagement curve keeps climbing)

The next month shifted toward a Boeing North Charleston hiring surge — Aircraft Systems, Electrical, and Structures Assembler campaigns. Smaller total outreach volume, tighter targeting, and a noticeable jump in engagement quality:

• 5,923 outreach messages sent
• 90.9% blended deliverability (up from 75.3%)
• 41.8% positive-intent rate (up from 26.5%)
• 24 applications started
• 4 referrals

Voice and chat usage in this window:

• 40 phone conversations handled by Hannah (≈3x prior period)
• 127 web chat conversations
• 82% engagement rate on phone
• 91% engagement rate on web chat

The phone line went from a curiosity to a real channel. Candidates started calling in after hours, and Hannah handled the intake instead of dropping them into voicemail.

What changed between the two months

The engagement curve did not climb on its own. A handful of specific product changes landed between the two periods, and they directly drove the jump in deliverability and positive-intent rates:

• Voice agents are now auto-provisioned for every new job, with candidate questions personalized to the job instead of a generic intake script
• Follow-up shortlists are auto-created from prior engagement, so the second-touch outreach only hits candidates who actually responded the first time
• Voice library refresh for a cleaner, more natural-sounding persona
• MS Teams support for recruiter handoff
• Progress tracker for campaigns in flight
• Multiple emails and phone contacts per candidate (because shipyard candidates change numbers constantly)

None of those are glamorous. They are the unglamorous plumbing that turns a pilot into a working workflow.

What one real conversation looked like

Aggregate numbers are easy to wave away. Here is a single real example.

A candidate we will call Garret was originally contacted via SMS for a Hawthorne machinist role. He did not reply right away. Weeks later, he came back and started a conversation with Hannah.

Over roughly 30 back-and-forth messages, the system captured:

• CNC machinist roles
• at least $35/hour
• contract work only
• initially local, then open to opportunities across the US

From there, the system surfaced a matching live req in Mesa, AZ, collected the qualification details, and got the application routed forward — without a recruiter ever stepping in until the handoff was warm.

No portal maze. No “create an account to apply.” No 45-minute back-and-forth to piece together pay, trade, and location.

That is the pattern repeating across the shipyard campaigns too: a Ship Fitter who engages after shift, tells Hannah they want fleet work, $38/hr minimum, open to traveling, and is surfaced into an active Winter Fleet req the same night.

Why shipyard staffing teams should care

The numbers are only interesting if they connect to real pipeline outcomes.

On the aerospace side of the same account, we ran an attribution analysis across roughly 190 unique candidate downstream outcomes (26 hires, 164 submittals). That analysis found:

• 85 candidates with documented TalentPrism activity across the submission set
• 28 candidates had documented engagement before the file or submission date
• 8 of 26 hires (31%) were touched by TalentPrism before hire
• 20 of 164 submittals (12%) had pre-file activity before submission

Translated out of metric-speak: roughly one in three hires on that campaign had an AI-driven chat, voice, or outreach touchpoint before the recruiter ever submitted them. That is not a chatbot vanity metric. That is pipeline influence on real placements.

The same workflow is already running against the shipyard campaigns. Same Hannah, same chatbot, same intake. Just pointed at Ship Fitters and Structural Welders instead of Aircraft Assemblers.

Three ways firms are actually using this

Across the accounts running this workflow, usage tends to fall into three concrete starting points. You do not have to buy all three at once:

1. Web concierge — Hannah lives on the careers page and replaces static apply forms with a real conversation. Best if you already have steady site traffic and mostly collect forms.
2. Phone receptionist — Hannah answers inbound candidate calls 24/7, captures interest, runs structured intake, and summarizes the conversation for the recruiter. Best if you are branch-heavy or bleeding candidate calls after hours.
3. Database reactivation sprint — point Hannah at one req family and one segment of your ATS, run outreach + follow-up, and surface updated availability. Best if you have 10+ years of records sitting mostly unused.

All three feed the same candidate memory layer, so the Ship Fitter who chats on the website, calls the line a week later, and responds to a reactivation text three months after that shows up as one compounding candidate record, not three disconnected pings.

What the operators are saying

Two quotes from inside the workflow:

“The chatbot is very intelligent. I love how it enables candidates to apply for a job right from the chat!” — Recruiter, after reviewing the widget against a live staffing workflow

“I love all the stats and analysis you are able to provide.” — Staffing firm owner, after reviewing the Boeing campaign reporting

The second quote matters more than it looks. Most staffing firms give up on chat and voice not because the candidates refuse to use them, but because they cannot see whether the tool is actually working. Once the reporting catches up, the argument changes.

Try Hannah yourself

If you want to see what this looks like from the candidate side, the fastest way is to call or text the live Pascagoula demo line:

(228) 471-0534

It is a real TalentPrism voice and SMS workflow configured for shipyard trades. Call it, ask about welder roles, tell it you are a Ship Fitter looking for fleet work, see how it handles the intake. Text it and let Hannah walk you through the same flow.

If it sparks ideas for your team, happy to hop on a short call and show how the same setup would run against your own candidate database — web, phone, or reactivation sprint, pick one to start.

The bottom line

Shipyard staffing does not need more software theater. It needs tools that capture candidate intent after hours, collect usable trade and availability data without burning recruiter time, and show up in actual submissions and hires.

Two months of real usage data says Hannah is doing exactly that — across Machinist, Marine Electrician, Structural Welder, and Ship Fitter campaigns on one side, and aircraft assembler campaigns on the other.

Not a chatbot. A 24/7 candidate activation layer for staffing firms.