Voice & Text-to-Execute
Tell Deptly once. Get the follow-up, update, task, and system work handled.
How a voice note from the truck becomes a sent estimate, an updated schedule, and a notified crew — with approvals wherever your rules require them.
What does this layer do?
Voice-to-execute means you can say what needs to happen — in a voice note from the truck or a two-line text — and Deptly turns it into finished work. It transcribes the note, matches names and addresses to real records in your business memory, extracts each distinct action, executes what your rules allow, queues the rest for your approval, and confirms exactly what was done.
Say it once, out loud
The gap between finishing a job and recording what comes next is where service businesses leak. You quoted a number in a driveway, promised a callback, agreed to move a job — and now that information lives nowhere but your memory, competing with the next four stops. The write-up, the CRM entry, the crew text: that is tonight's work, if it survives until tonight.
Voice-to-execute closes that gap at the moment it opens. You say what happened and what needs to happen — messy, out of order, half sentences — and the system work becomes Deptly's job instead of your evening.
How a messy note becomes structured work
First, speech-to-text transcription turns the audio into words. That part is commodity technology; the useful work starts after. Deptly reads the transcript for meaning: which statements are instructions, which are context, and how many distinct actions are packed into one rambling note.
Then comes entity matching — connecting the words to your actual records. The Hendersons resolves to a specific customer and property in your business memory. The gray house next door becomes a new lead with a real address. Marco is your crew lead, and the Fairview job is a specific entry on a specific calendar. This grounding step is what separates executing your note from merely transcribing it.
Finally, multi-action extraction splits the note into its separate jobs — write the estimate, create the lead, move the job, tell the crew — each routed down its own path with nothing dropped because it was mentioned in passing.
Approvals and confirmation
Extracted actions follow the same control rules as everything else Deptly does. Routine work inside your approved rules executes directly. Anything outside them — a price, a discount, a commitment — queues for your yes before it goes anywhere. Saying something out loud does not bypass your own controls.
Ambiguity becomes a question, not a guess. Two customers named Miller, an address that will not resolve, a number that could be $620 or $6,200 — Deptly asks. And when the work is done, you get a confirmation of what was understood and executed, so a misheard note surfaces immediately instead of three days later as a wrong estimate.
Where it works
A voice note sent by text. A rambling message to your business line. A forwarded email with just handle this. Your team can use it too — a crew lead reporting a finished job with photos, an office manager relaying a change — each within the permissions you have set for them.
There is no app to learn and no form to fill out in a parking lot. The input is however the thought naturally comes out; the structure is Deptly's job.
What it works from
- Voice notes and texts from you or your team, in whatever state they arrive
- Speech-to-text transcription as the first step, with meaning extraction on top
- Business memory, to match names, addresses, and jobs to real records
- Your calendar and systems, to check whether an instructed change actually fits
- Your approval rules, to decide what executes now and what waits for a yes
Everything above comes from your business — your records, your rules, your tools. Nothing is invented, and when information is missing, it becomes a question to you rather than a guess.
What this looks like in practice
Every proof item on this site is labeled for exactly what it is.
One voice note, three actions closed — landscaping
Owner (voice note)
Just left the Hendersons — quoted them verbally at sixty-two hundred for the patio extension, need that written up. Their neighbor at the gray house wants a cleanup quote too. And tell Marco the Fairview job moves to Friday.
4:52 PM, from the truck
Deptly
Matched: Henderson property (patio project, $6,200), new lead at 41 Birchwood (cleanup request), Fairview Ave job, crew lead Marco. Three actions extracted.
4:53 PM
Deptly
Needs your OKHenderson estimate drafted at $6,200 from your pricing records — approve to send? Fairview's move to Friday has no conflicts; Marco gets notified once you confirm.
4:55 PM
Owner
Send it. Confirm Friday.
4:58 PM
Deptly
Estimate sent to the Hendersons, Fairview rescheduled to Friday in the calendar, Marco texted the change, and the Birchwood cleanup lead is created with a quote visit to schedule. All three items closed.
5:04 PM
Where humans stay in the loop
This layer runs inside boundaries, not on its own. Here is who reviews what — and the limits it never crosses.
Human review
- Deptly confirms what it understood — extracted actions are shown as they are executed, so a misheard note cannot quietly become wrong work.
- Ambiguous matches — two customers with the same name, an address it cannot place — become questions, not guesses.
- Actions outside your approved rules, like a new discount or a schedule exception, queue for your explicit approval.
- A voice note can never authorize a legal, clinical, safety, coverage, or fiduciary decision — those route to a person every time.
Guardrails
- Nothing in a note is silently dropped — every extracted item is executed, queued for approval, or explicitly flagged as unclear.
- Money, commitments, and exceptions in a note follow the same approval rules as any other action — saying it out loud does not bypass your controls.
- If a note contradicts recorded policy, Deptly asks rather than assuming the note wins.
- Recordings and transcripts are retained under your data settings and are yours to export or delete.
The full rulebook — control levels, approvals, and audit history — lives on trust and control.
Where this layer shows up
Technology is only interesting when it carries real work. These are the responsibilities this layer powers — and the layers around it.
Responsibilities it powers
The layers around this one
Questions owners ask
Two safety nets catch it. Entity matching checks the transcript against your real records — a name or address that does not resolve becomes a question, not a guess. And you see a confirmation of what was understood and done, so an error surfaces immediately. Anything involving money or commitments also passes through your normal approval rules before it goes out.
By matching the note against business memory — your customers, properties, jobs, and staff. "The Hendersons" resolves to a specific record; "the gray house next door" becomes a real address from context. When more than one record plausibly matches, Deptly asks you to pick rather than choosing for you.
It accounts for everything, but it does not freely act on everything. Each extracted action is either executed under your approved rules, queued for your approval, or flagged as unclear. The guarantee is completeness — nothing mentioned gets dropped — not that speaking a thing makes it happen without your controls.
Voice notes and texts to your business line, and forwarded emails, are the core channels. The practical answer for your specific phones and tools gets settled during onboarding — the design intent is that you use whatever you already reach for in the truck, not a new app.
Yes, within permissions you set. A crew lead might report job completion and request materials; an office manager might relay schedule changes. Approval authority stays mapped to real roles — a crew note can trigger drafts and questions, but it cannot approve a discount or waive a policy unless you have given that person that authority.
Recordings and transcripts are kept under access controls and your retention settings, and they are yours — you can export or delete them. During onboarding we walk through exactly what is retained and for how long for your setup, rather than making one-size-fits-all claims here.
Deptly asks instead of assuming the note wins. If you say offer them 15 percent off and your recorded ceiling is 10, that comes back as a quick check: intentional exception, or slip? Your answer settles it — and if it is an intentional exception, it gets logged as one instead of quietly rewriting your policy.
Under the hood, then in practice
See what Deptly could take off your plate
Answer a few questions about how your week actually runs and get an honest read on which responsibility this machinery should carry for you first.