ChatGPT or an intake tool for job orders?

Published 17 October 2026. Updated 17 October 2026.

A client sends three paragraphs about a role they need filled. You want a clean requisition, a list of what is missing and a short reply to the client. A general AI chat assistant such as ChatGPT can read that email, so the question is fair: do you need a separate tool at all? This guide compares the two approaches on the points that matter in a staffing firm, says when the chat assistant is enough, and gives you a checklist that works whichever you pick.

What the job actually takes

Turning a job-order email into something your recruiters can work from is five small jobs, not one:

1. Pull out the fields. Title, client, location, pay or bill rate, headcount, must-have skills, nice-to-haves, dates, who the candidates report to.
2. Keep them apart. A nice-to-have must not become a hard requirement, and a pay range must not become a single number.
3. Check against your list. Your firm has fields it will not start a search without. Is each one present?
4. Ask about the gaps. A short reply that asks only about what is missing or unclear.
5. Match against open orders. Is this a new role or a change to one you already have? The guide on telling an update from a new requisition covers that decision in detail.

A chat assistant can help with the first, second and fourth by default. The third and fifth need something extra from you: your field list and your current open orders, which you have to supply and keep up to date.

Where a chat assistant works well

Be fair to it. For a small number of orders, a chat assistant is a sensible choice:

• It costs little to try. You can test it on five old emails this afternoon.
• It is flexible. You can ask it to restate the email as a table, shorten it, or draft the reply in your firm's tone.
• It handles odd emails. A forwarded chain, a long signature, a client who writes in fragments: all fine to paste and ask about.
• You control the prompt. If your process is unusual, you can describe it in your own words.

If you get a few orders a week, you already read each email closely, and you check the output against it every time, a chat assistant plus a good saved prompt may be all you need.

Where it breaks down

These are risks of the approach, not claims about any one product. Test each one yourself with your own emails.

• A plausible value for something the email never said. Language models write fluent text. If the email gives no start date, a model may still produce one that sounds right. Ask for blanks to stay blank, then test it on an email with a field deliberately missing.
• No record of where a value came from. A tidy table gives you no line to point to. If a recruiter asks "where did that rate come from?", you search the email again, which is the work you wanted to skip.
• Your required fields live in your head. Unless you give it the list, in the prompt or in saved instructions, nothing checks against it. It cannot know that your firm never starts a search without a fee agreement reference.
• It only knows the open orders you give it. It cannot say "this is an update to the Senior Accountant order" unless you supply your current open orders, and that list goes stale as orders open and close.
• The layout drifts. Column names, ordering and wording can change between runs, which matters the moment you paste into ATS fields or a spreadsheet.
• Checking is up to you. Whether the output is right is down to your own read, every time. That is fine on a quiet day and easy to skip on a busy one.

Side by side

When the chat assistant is the better choice

Choose it when your volume is low, when every order is different enough that a fixed layout would not help, or when you want one tool for many jobs besides intake. Choose it also if you enjoy refining prompts and will keep reading each result against the email. A careful reader with a good checklist and a saved prompt does this job well.

A single-task tool earns its place when orders arrive often, when more than one recruiter handles intake and you want the same fields every time, or when "did anyone check the second time?" is a question you cannot answer today.

A checklist that works with either

Use this whichever you pick:

1. Keep your required-field list in one file and use it every time.
2. Tell the assistant or tool that a missing value must stay blank and be flagged.
3. Keep must-haves and nice-to-haves in separate fields.
4. Keep pay as a range with its period (hour, month or year), as the client wrote it.
5. For each value, be able to point to the line in the email.
6. Compare with your open orders before creating a record.
7. Read the reply to the client yourself before you send it.

If you do not yet have a firm list of required fields, start with the fields every job order form needs. If the intake call is where details go missing, see the questions to ask on an intake call.

How ReqDrop does this step

ReqDrop is a single-task tool for the email-to-requisition step. You upload the client's job-order email, or paste a short one next to an uploaded file. You can add your own required-field list as a .csv or .txt file, one field name per line, and your open-requisition list as a spreadsheet, a Word file, a PDF or a photo. It reads the first 40,000 characters of the email and takes up to four files of 10 MB each.

You get back a checked job order as JSON and CSV (title, client, location, pay, headcount, skills, dates), a checklist in which each value is linked to its email line, and a reply to the client that asks only about the gaps. A missing value stays blank and becomes a question to the client. If you uploaded open orders, it flags an update to an open order instead of a new draft.

What it does not do: it does not write into your ATS, so you read the draft and paste it in yourself. It does not send the reply, so you edit and send it. It reads emails only, not phone calls or voicemail.

A second, stronger AI model reviews every result, fixes what it can, and lists what is left for you to decide. Results are model output checked by code and by a second model; they are not verified by a person.

One job order is 12 credits, from $8.40 to $12 depending on the pack, with no subscription. Sign up and try it on one of your own emails.

Questions

Can ChatGPT write the reply asking the client for missing details?

A general chat assistant can draft a reply if you tell it which fields are missing. The risk is the step before: it has to decide what is missing, and you only learn whether it got that right by checking the email yourself.

Is it safe to paste a client's email into a chat assistant?

That depends on the tool's data terms and on what your client contract says about sharing their information. Read both before you paste anything, and ask the client if the contract is silent.

Should I save my prompt so I get the same layout every time?

Yes. Keep one prompt that names your required fields, the output layout and the rule for blanks, and paste it with every email. If the wording changes between runs, the layout can change too, and the output stops matching your ATS fields.