Draft a Customer Project Update With AI at the Cursor

The short answer: speak four verified facts (completed work, remaining task, dependency, and next date) together with a clear instruction, and let a spoken AI request produce a structured draft at the cursor in a compatible Windows text field, such as an open email or message. Then review every line yourself before anything is sent. This suits small-business owners who know the status of a job but lose momentum in front of an empty reply box. It works only where the text field and the service are compatible, and the draft never replaces your own check of the facts and promises in it.

Below you will find what a customer update should contain, how to speak the inputs, a before-and-after example, a reusable template, a review checklist, and a method for measuring time to reviewed output from your own results.

A project update has a narrow job: tell the customer where things stand, what happens next, and what, if anything, you need from them. When owners write these at the end of a long day, the common problems are weak structure and vague wording. A long paragraph buries the request, "soon" stands in for a date, or a friendly line reads like a commitment you never made.

Four facts carry most of the message:

• Completed work: what is finished, in terms the customer would recognise.
• Remaining task: what is still to be done.
• Dependency: anything the work waits on, such as the customer's decision, a supplier, access to the site, or an approval.
• Next date: the next concrete date or checkpoint, included only if you actually know it.

Once you hold those four facts, the remainder is structure and tone. An AI-assisted draft can help with those two things. The facts and the promises stay with you.

SpeakLingo is built around Syping, a way of working that combines speaking and typing. You type what is quick to type and speak when explaining a situation aloud is easier than composing it. The idea is explained further on the Syping page.

Three capabilities matter for a customer update:

• Online transcription in 99+ languages, so you can speak your notes in a supported language that suits you and your customer relationship.
• AI contextual formatting, which can shape spoken input into a structured layout, such as a short greeting, labelled sections, and a closing line.
• Spoken AI requests that produce text at a compatible, focused Windows text field. If your cursor is in the reply box, the text is produced there. Where this workflow fits, you do not need a separate AI window and a copy-and-paste step into your email.

Some boundaries apply. SpeakLingo is not compatible with every application or field. The workflow is meant to leave a draft at your cursor for you to review, but an optional Enter-after-delivery behavior can send a message in some apps, so keep it off for customer updates. AI requests need an internet connection, an enabled service, your consent, and access. Audio and text may be sent to remote services to process a request, and optional screen context may be sent if you opt in. All AI output is a draft that needs your review.

1. Set up for review, not for sending. Before your first request, turn off any automatic Enter or send behavior, so that delivered text lands in the field and waits for you. The draft should sit there until you have read it.
2. Gather the facts before you speak. Open your job notes, calendar, or supplier confirmation and verify each of the four inputs. If you do not know one, plan to say so aloud instead of guessing.
3. Place your cursor. Open the reply or new message and click into the body so the field has focus. If the field turns out to be incompatible, you will find out early and can write the message your usual way.
4. Speak the instruction and the inputs. Use a pattern like the one below, which tells the AI to use only what you provide and to flag gaps.
5. Read the draft against your notes. Compare each sentence to the four facts and delete anything you did not say.
6. Resolve every flag. If the draft lists a missing fact, supply it from a verified source or remove the sentence that depends on it.
7. Edit for voice, then send it yourself. Adjust the greeting and tone to match how you normally write to this customer, and press send only when you are satisfied.

The request pattern, with the bracketed parts replaced by your own facts:

"Using only these verified inputs, draft a customer project update. Inputs: [Completed work, remaining task, dependency, next date]. Use a clear professional structure. Flag missing facts; do not invent them."

The pattern contains two guardrails: use only the inputs given, and flag what is missing. They ask for careful behaviour and cannot guarantee it, which is why the review step stays in place.

One boundary matters for a business owner who is busy and speaking quickly. Dictating your notes as ordinary transcription does not switch AI on. To get a drafted update, use the supported AI Task workflow and give the instruction and the job facts together in that one request. Your earlier dictations, any text already in the reply box, the email thread, and whatever is on your screen are not automatically available to the AI, so say the facts you want used. Screen context is a separate opt-in, and it may be sent to remote services. AI requests need an internet connection, an enabled service, your consent, and access.

The scenario below is fictional, with an invented customer and business. The draft is illustrative output written to show the format. It is not observed output from a test of the app. The example shows how scattered notes become a reviewable draft.

Situation: A small home-services business is tiling a kitchen backsplash for a customer named Dana. The owner has been on site all day and needs to send an update.

Before: the owner's rough notes

"Tile's done on the backsplash. Still got grouting and the sealing. Need Dana to pick the grout colour, she said she'd think about it. Probably back Thursday or so. Mention the countertop thing? Not sure. Keep it friendly."

Sending this as written would cause several problems. "Probably Thursday or so" reads like a loose promise. The dependency is buried mid-note. The countertop remark is unexplained, and it is unclear whether it belongs in this update at all.

The spoken request at the cursor

The owner clicks into the email body and speaks:

"Using only these verified inputs, draft a customer project update to Dana. Completed work: backsplash tile installed. Remaining task: grouting and sealing. Dependency: Dana confirms grout colour. Next date: not confirmed; Thursday is possible but do not state it as a firm date. Leave out the countertop and do not mention it. Use a clear professional structure. Flag missing facts; do not invent them."

After: the draft at the cursor (illustrative output)

Subject: Backsplash update and one decision needed

Hi Dana,

Here is where your kitchen backsplash stands.

Completed: The backsplash tile is installed.

Still to do: Grouting and sealing.

What we need from you: Please confirm your grout colour so we can proceed.

Next date: We may be able to return on Thursday, but that date is not yet confirmed.

Let me know if you have any questions.

Kind regards, [Name]

Flags for the sender (delete before sending): - No deadline was given for the grout colour confirmation. - Thursday is not confirmed, and it is unclear whether it depends on the colour being confirmed.

The owner's review

The owner reads the draft against the notes and makes four decisions:

• Thursday and the dependency: The owner checks the schedule, confirms Thursday is workable, and decides the Thursday visit works only if Dana confirms by Tuesday. That is a business decision, so the owner types it in: "If you can confirm by Tuesday, we can grout on Thursday."
• The countertop: The owner had already decided it needs its own conversation, so the request told the AI to leave it out. The owner confirms the draft does not mention it.
• The tone: The owner replaces "Let me know if you have any questions" with a closing line they usually use with Dana.
• The flags: Each one is resolved or deliberately dropped, and the flag block is deleted.

The final message contains only facts the owner confirmed, and the owner presses send. The Tuesday deadline was the owner's decision, and the countertop stayed out because the owner chose to keep it for a separate conversation. Both calls belong to the person who knows the customer and the job.

Keep this wording somewhere handy and adapt the bracketed parts. You can speak it or type it.

Using only these verified inputs, draft a customer project update to [customer name].

Completed work: [what is finished, in the customer's terms] Remaining task: [what is left] Dependency: [what the work is waiting on, and who owns it] Next date: [date or checkpoint, or "none confirmed"]

Use a clear professional structure with a short greeting, labelled sections, and a brief close. Keep the tone [friendly / formal / brief]. Flag missing facts. Do not invent them. Do not add promises, prices, outcomes, or dates I did not give.

Two habits sharpen this template. Say "none confirmed" when you have no date, which gives the AI an explicit gap to flag. And keep each input short and factual, because a long, rambling input leaves more room for loose wording and gives you more to check line by line.

The table treats each field as a small writing decision. These are illustrative writing practices for a project update, and they do not amount to contract, legal, or professional advice.

Work through this list on every draft, in this order:

• Facts: Does each statement match a source in front of you, such as job notes, a supplier message, or your calendar?
• Invented content: Is there any outcome, benefit, reassurance, or result that you did not say? Remove it.
• Promises: Does any sentence commit you to a date, a price, a quality level, or a result? Keep only what you are willing to stand behind.
• Dependency clarity: Is it obvious what you need from the customer and, where relevant, by when?
• Dates: Are all dates and weekdays correct, and does the next date still hold if the dependency is delayed?
• Names and details: Are the customer's name, the address or job reference, and the item names correct?
• Flags: Has every flagged gap been resolved, or has the related sentence been deleted?
• Sensitive details: Does the draft include anything you would not want in this message, such as another customer's information or private financial detail?
• Tone: Does it sound like something you would send under your own name?
• Send decision: Is the message ready on its facts and promises, whatever the polish of the wording?

If you opt in to screen context, also check what is visible on screen when you make the request, since that context may be sent remotely. Close windows that show other customers' details before you speak.

Your own records will tell you more than any general claim about speed. The useful measure is time to reviewed output: the time from the moment you start preparing the update to the moment you hold a version you have checked and approved for sending.

A simple way to measure it:

1. Define the start and end. Start when you begin gathering facts or speaking the request. End when your review is finished and you are ready to press send. Leave out any time the message sits waiting afterward.
2. Record several updates. A single update can be unusually easy or hard, so log a handful for each method.
3. Compare like with like. Match routine status updates with other routine ones, and complicated updates with complicated ones.
4. Track edits alongside time. Note how many sentences you rewrote or deleted and how many flags you had to resolve. A quick draft that needs heavy repair costs differently from a slower one that needs little.
5. Note the fit. Record any occasion when the text field was incompatible or the service was unavailable, since those occasions are part of your real workflow.
6. Decide from your own results. If the records show no benefit for a certain kind of update, keep writing those by hand.

Online transcription supports 99+ languages, so you can speak in a supported language that suits you and your customer. That statement concerns transcription only. SpeakLingo is not described here as a translation tool, and AI-formatted output is not guaranteed to work equally well in every language. If you write to customers in a language other than English, read the draft with extra care, since you are the one who can judge how it will sound to the recipient.

AI requests require an internet connection. If the connection is interrupted, AI requests are unavailable. Offline English transcription is only a fallback, and only if it has been configured beforehand; it captures your words but does not provide AI drafting. Transcription has its own mode, setup, and access requirements, so an AI outage does not show that transcription is available. In that situation you would write the update yourself from the same four facts, and the template above still works as a checklist for your own drafting.