Qelv

Back to Services

Automate

We make the repeat work run by itself.

Work that happens on a schedule, or on an event, without a person driving it: agents that answer and qualify, documents that get read and filed, systems that exchange records directly, reports that assemble themselves.

Projects
21

6 written up on the work page

Figures published
8

listed on the work page

Free first step
Automation Map

built from the process eating the most hours, or the role you are about to post

Fixed price from
$1.5K

Hire-Deflection Automation Sprint

Is this you

The week this fixes.

Sentences we hear from the people who end up sending this practice a bottleneck, and what each one is costing while it stays true.

  • “We are about to hire someone to do all of this.”

    A salary spent on work that is mostly rules, not judgment.

  • “Someone has to copy that into the other system.”

    A person with two tabs open, retyping one system into the other.

  • “The Monday report takes someone the whole morning.”

    Numbers pulled out of three systems every week and pasted into the same sheet.

  • “A form comes in at 7pm on Friday. Someone sees it Monday.”

    Customers waiting on whoever opens the inbox next.

That is us. Send the bottleneck

Or start with the free Automation Map, below.

What a project covers

Scoped to the smallest useful system, judged against a number.

  • Agents that handle a whole interaction: voice, chat or email, with escalation to a person where it belongs
  • Document and data processing pipelines: extraction, classification, routing and filing
  • Workflow automation between the systems you already run
  • Monitoring, so an automation that breaks tells someone instead of silently stopping

What that has looked like

  • A voice or chat agent that answers, qualifies, books, and writes the outcome back into the business system
  • Document intake that reads, classifies and files what arrives, at volume
  • Two systems exchanging records directly instead of through a person

The measures we agree before building

One of these is picked with you at the scoping step, and reported against.

  • Hours a week returned to the team
  • Cost per processed item (order, ticket, invoice, application)
  • Cycle time from request to done, and the error rate along the way
Send the bottleneck for an Automation Map

We reply within two business days. If it is a fit, you get a one-page note on what to do first, before any sales call.

Your own arithmetic

What is the repeat work costing already?

Your numbers, not ours. Move the sliders to whatever is true for your team.

On your numbers, every year

552 hours

12 hours a week across 2 people, over 46 working weeks.

$16,560

At $30 an hour, fully loaded.

An automation usually takes over the rules-based part of this work, not all of it. We tell you which part before quoting anything.

Send these numbers with the bottleneck

Your numbers travel with the message and reach us when you submit the form. If your browser blocks site storage, as a private window usually does, the form opens without them and you can type them in. They are kept out of the web address either way, so they are not in your history and nothing we send to analytics carries them. Clicking records only that the estimator was used.

No charge, no call

Start with the free one.

Each is a document or a sample, written by the person who would do the work. Send one input, get something back you can act on or ignore.

  • Free

    Automation Map

    Every task in a role sorted into three columns: software can do this, a person has to, not worth touching either way.

    You send: The job posting you are about to run, or a description of the process eating the most hours.

    You get back:

    • The posting marked up task by task, in three columns
    • Which of the software column is worth doing first, and roughly what it costs
    • What the seat still needs a person for, so the hire starts further along

    Two business days. About twenty minutes to write from the posting itself. No call attached, and no obligation to stop hiring.

    Usually precedes Hire-Deflection Automation Sprint. Nothing obliges you to take it.

Every free first step across the five practices is listed together, if the problem sits between two of them.

Entry points

The ways in, each priced before work starts.

Start with one. None of them requires the next one, and each is small enough to judge on its own.

  • Hire-Deflection Automation Sprint

    Automate the repeatable part of an open ops, admin, marketing, or support role before filling the whole seat.

    $1.5K–$3.5K · Fixed price · optional care plan afterwards · 2 to 3 weeks

  • The Monday Report

    The weekly report builds itself, out of the systems the numbers already live in, and lands before anyone asks for it.

    $1K–$2.5K · Fixed price · 1 to 2 weeks

  • System-to-System Sync

    Data entered once. Your CRM and accounting stop needing a person to copy between them, and it keeps running after we are gone.

    $1.5K–$3K · Fixed price · 1 to 2 weeks

  • Intake and Inbox Agent

    Every inbound form, email and quote request answered and routed the same hour, with anything it cannot judge escalated to a person visibly.

    $1.5K–$3K · Fixed price · 2 to 3 weeks

Every price, fixed or by the hour, is agreed before work starts and built the same way: the hours, the tools, and a share of company cost. See how a price is built

The free first step

What the first move looks like.

Examples of what we sent back after a bottleneck came in.

Between September 2025 and March 2026 we held about forty-five discovery calls, each used to define a practical first move.

Automation Map, delivered as a system diagramOutreach, inbox, follow-up, voice, and billing, mapped as one systemWhat happened: Proposal only. No signature, no build.

A services company wanted its outreach, inbox triage, follow-up, voice calls, and billing to stop living in five tools that a person bridged by hand.

What it covered

  • A diagram of the whole system across an automation platform, the CRM, a voice layer, and payment and accounting
  • Acceptance criteria written as measurable targets, so the build could be judged rather than admired
  • A two-phase plan, with the second phase contingent on the first being trusted
The first move
Automate intake and follow-up first; voice and billing wait until the classification is trusted.

2025

Automation MapBilling reconciliation for a small clinical practiceWhat happened: Proposal prepared; no award recorded.

A psychology practice reconciled its billing by hand every month, and had a budget that ruled out a platform.

What it covered

  • The monthly reconciliation as it actually ran, step by step
  • The hours it cost, in the client's own estimate
  • What a small automation would take over, and what should stay manual
The first move
One small automation inside the budget, priced as a build rather than a subscription.

September 2025

Automation Map, delivered as a two-phase proposalLead intake, qualification, follow-up, and booking, in two phasesWhat happened: Proposed; no award.

A company wanted inbound leads qualified and followed up without a person watching the inbox, and booked when they were ready.

What it covered

  • A workflow-platform design for intake, qualification rules, follow-up sequences, and booking
  • Phase one and phase two, with the second only after the first had run
  • An hour estimate rather than a fixed price at that stage, and the reason why
The first move
Phase one: intake and qualification only.

March 2026

Automation Map, with a statement of workGuest-service automation for a hospitality operatorWhat happened: Proposed.

A hospitality operator wanted routine guest requests handled automatically, without the guest noticing a machine.

What it covered

  • The requests that could be automated, and the ones that need a person on the first reply
  • The escalation path, written before the automation
  • A statement of work scoped to the routine requests only
The first move
Automate the routine requests; route everything else to a person, visibly.

November 2025

Demo-first discoveryA spreadsheet-validation demo built before the proposalWhat happened: Demo and proposal; no award recorded.

A manufacturing prospect asked for a recorded demo of the tool. There was no tool: the work was custom, so a working front end was built to show what it would do.

What it covered

  • A front end that takes a spreadsheet upload, runs validation rules (an empty product code, a duplicate identifier), and hands off to workflow templates
  • An indicative proposal with two options and a work breakdown by activity
  • An explicit caveat that requirements had been gathered by email and no formal discovery had happened
The first move
Something to click before anything to sign.

February to March 2026

Send us the bottleneck

The note comes back before any sales call, or we say it is not a fit.

The work

What this looks like shipped.

A high-volume hotel pricing-data system

169 properties, same day

Delivery record, Qelv engineering record · 2026

Scheduled collection, proxy rotation, retry, and anti-blocking strategy, and the operational work of keeping a high-volume pipeline healthy, without adding headcount to maintain the feed.

In production

4 figures

Under client agreement. The client owns this system. Ask and we will walk you through how it was built and what it changed.

Request a walkthrough

A retail catalog pipeline that keeps a Shopify store current

Scrapes Costco product data on a schedule, normalizes and stores it, and syncs price and stock to Shopify.

Our engineering · Delivered

Read the case study

A café operations system: orders, inventory, shifts, loyalty

20-table schema

Qelv engineering record · Built March 2026

An internal product built in March 2026: point of sale, the barista queue, an inventory ledger, recipes, loyalty, shifts, and void reversals across a 20-table schema, designed against a 22-entity domain model.

Internal product · Delivered

1 figure

Internal product. Built and operated by us, not sold as a product. Happy to demo it live and explain the decisions inside it.

Request a live demo

Public-records search and retrieval, automated

Drives the Hamilton County clerk-of-court foreclosure search with Selenium and retrieves each matching PDF, retrying failed downloads: a job otherwise done by hand, one case at a time.

Delivered

Source not public. Our own engineering. We will screen-share the code, the architecture, and the tests with anyone seriously evaluating us; publishing the repository invites the scrapers we spend our days defeating.

Request a code walkthrough

Google Sheets to PDF workflow automation

A small automation, built in August 2023, that keeps Google Sheets as the working interface, reads the worksheet records through the Sheets API with gspread, and writes structured PDF output with FPDF.

Our engineering · Delivered

Source not public. Our own engineering. We will screen-share the code, the architecture, and the tests with anyone seriously evaluating us; publishing the repository invites the scrapers we spend our days defeating.

Request a code walkthrough

A cold-outreach engine we ran on our own pipeline first

247 qualified leads

Email outreach results report, Qelv internal record · July to October 2025

Contact sourcing from Apollo, verification, warmed secondary sending domains, sequenced copy, reply triage, and weekly deliverability reporting. We ran it at scale on our own program through the second half of 2025, then for a partner under their brand, as a pilot for a client, and on hourly contracts.

Internal product · Delivered

3 figures

Read the case study

6 more automation projects

What else we have built in this area, and what changed as a result.

Engagements in this catalogue, described without numbers and, in most cases, without the client: they were built under confidentiality or under another firm's brand. For numbers with sources behind them, see the work page.

OperationsSupplier invoices, read and filed without a person

Invoices arrived as PDF attachments from dozens of suppliers, each in its own layout, and were keyed into the accounting system by hand.

What we built

  • Mail-drop intake that picks up attachments and classifies them by supplier
  • Field extraction per supplier layout, with a confidence threshold
  • A review queue for anything below the threshold, so nothing is silently wrong
  • Posting into the accounting system with the original document attached

What changed

Keying invoices became reviewing exceptions, and month-end stopped being a backlog-clearing exercise.

Fixed-price sprint, then a monthly care plan

Python · Document parsing · Accounting API · Queue worker

Under client agreement. The client owns this system. Ask and we will walk you through how it was built and what it changed.

Request a walkthrough

OperationsA CRM that stopped needing to be tidied

Duplicate records, stale ownership, and half-filled fields meant reporting was rebuilt by hand every month and nobody trusted the pipeline number.

What we built

  • Deduplication and merge rules that run continuously rather than in an annual clean-up
  • Enrichment from the systems that already knew the answer
  • Validation at write time, so bad records stop being created
  • A standing report built from the CRM instead of from a spreadsheet export

What changed

The monthly clean-up stopped being scheduled work, and the pipeline number became one people quoted without checking.

Fixed-price sprint

Python · CRM APIs · Scheduled jobs · Reporting

Under client agreement. The client owns this system. Ask and we will walk you through how it was built and what it changed.

Request a walkthrough

OperationsRota building, from a whole afternoon to a review

A manager spent an afternoon a week building a rota in a spreadsheet, working around availability, certifications, and hours caps held mostly in their head.

What we built

  • Constraints written down as rules: availability, qualifications, maximum hours, rest gaps
  • A generated draft rota that satisfies the hard rules and flags the soft ones
  • Manual override kept first-class, because the manager knows things the rules do not
  • Publication and change notification to staff

What changed

Rota building became reviewing a draft, and the rules stopped living only in one manager's memory.

Fixed-price build

Python · Constraint solving · Next.js · Notifications

Under client agreement. The client owns this system. Ask and we will walk you through how it was built and what it changed.

Request a walkthrough

OperationsAudit evidence collected continuously instead of in a panic

Every audit cycle began with weeks of screenshotting settings and chasing colleagues for confirmations that a control was in place.

What we built

  • Scheduled collection of the evidence the auditor asks for, from the systems that hold it
  • Timestamped storage, so evidence is dated rather than reconstructed
  • Exception alerts when a control drifts, rather than discovery at audit time
  • An export in the shape the auditor already expects

What changed

Audit preparation stopped being a project, and control drift surfaced when it happened rather than months later.

Fixed-price build, then a monthly run

Python · Platform APIs · Object storage · Alerting

Under client agreement. The client owns this system. Ask and we will walk you through how it was built and what it changed.

Request a walkthrough

EcommerceA product business paying for tools it had stopped using

Ecommerce

An ecommerce operation had accumulated overlapping subscriptions and manual steps, each added to solve a problem that had since moved.

What we built

  • A map of what every tool and manual step was actually doing
  • Consolidation onto the systems that earned their place
  • Automation of the hand-offs that had been holding the stack together
  • Cancellation of what was left, with the workflow proven first

What changed

The operation ran on fewer moving parts, and the remaining ones were connected rather than bridged by a person.

Fixed-price sprint

Ecommerce platform · Automation platform · Integrations

Under client agreement. The client owns this system. Ask and we will walk you through how it was built and what it changed.

Request a walkthrough

OperationsTwo workshops on a cash-flow forecast assembled by hand every week

Logistics

A weekly treasury forecast was built by hand from two reports out of a freight-management system, by a team in one country for a sponsor in another, and nobody had written down how it actually worked.

What we built

  • A discovery call and a written intent summary agreed with the sponsor before any workshop ran
  • Two structured workshops walking the receivables and payables reports the forecast is built from
  • A question list and a bilingual non-disclosure agreement, with a Spanish-language session arranged, because the people who do the work and the people who fund it did not share a first language
  • A written discovery output naming what could be automated and what could not

What changed

The forecast's real inputs were written down for the first time, and the decision about whether to build was taken against them rather than against a demonstration.

Two workshops, delivered before any build was proposed

Discovery · Freight-management reporting · Process mapping

Client deliverable. The deliverable belongs to the client: the documents, the architecture, and the planning stay theirs. What we can walk you through is the method and the shape of the output.

Ask about this work

Browse the full catalogue

Where to next

The free first step: an Automation Map.

Send the process eating the most hours, or the role you are about to post.

We reply within two business days. If it is a fit, you get a one-page note on what to do first, before any sales call.