yourprivatedev
Book a call

Servicesix builds to show

Back-office automation and custom AI agents for operations and finance teams

Your Private Dev builds back-office automation and custom AI agents for the operations and finance teams of established small and mid-size businesses. Shipped work includes an agent that turns contractor emails into finalized Business Central invoices (70+ hours a month removed), a pipeline that produces annual member-club reports end to end (250+ hours a year removed), a two-way Brightpearl and HubSpot sync, MCP agents that let a roofing contractor's owner run operations from one chat window, and an AI discovery platform that extracted an 83,000-page case at about $0.003 per page. One senior engineer does the building, on a fixed-scope written quote.

Book a 15-min call See the case studies

Custom AI agents that finish the task

An agent reads the request, checks your systems and carries the task through. Kintec's agent, Dave, parses free-form contractor requests that arrive by email, cross-verifies the data against Business Central through its REST and OData endpoints, generates the invoice with the correct posting logic, validates it against accounting rules and historical patterns, and finalizes the payment workflow. Standard invoice cycles run with no human in the loop.

For a roofing contractor, the owner and ops team were running the business through five disconnected tools: CRM, scheduling, dispatch, invoicing and comms. Each operation was wrapped as a tool and exposed to the agents through MCP (Model Context Protocol), with a multi-agent layer that routes each request to the right tools. The owner now asks in plain language, for example to schedule an inspection or to list everyone whose deposit has not cleared this week.

Built for Kintec Global · Innovative Roofing Pros

Automation for work your team repeats every month or year

Distinguished Clubs generates a detailed Word evaluation report for every member club every year. The pipeline aggregates the data across internal APIs, surveys and club records, maps the mail-merge fields, runs the merge, auto-fixes the recurring layout issues and publishes the finished report, removing 250+ hours a year of manual work.

PlanYear's benefits consultants needed structured data from a portal across medical, dental, vision, life and more. A Playwright-based extractor logs in, navigates every plan type, handles the portal's quirks and outputs clean, normalized datasets, so consultants no longer copy plan data into spreadsheets by hand.

Built for Distinguished Clubs · PlanYear

Keeping your ERP and your CRM in step

Aid Through Trade runs orders, customers and its product catalog in Brightpearl while sales works in HubSpot. HubSpot changes reach Brightpearl in real time and Brightpearl data reaches HubSpot daily, with loop protection so a note synced one way never comes back as a duplicate. Orders show in HubSpot as deals with line items, so sales sees what each customer buys.

Built for Aid Through Trade

AI that works on your documents, with a source for every answer

CrossLogic needed a platform that reads every file in a criminal case. It ingests documents, emails and recordings, extracts people, organizations, locations and dated events into one timeline, and answers questions through hybrid retrieval with every answer citing the document it came from. AI spend is metered per case and stage: an 83,000-page federal case was fully extracted at about $0.003 per page in AI cost.

Built for CrossLogic Intelligence

Six back-office builds, with the detail

  • 70+ hof manual invoicing removed every month

    Kintec Global · 2026. An email-driven agent that checks Business Central, invoices and finalizes: 70+ hours a month of manual invoicing removed.

  • 250+ hof report production removed every year

    Distinguished Clubs · 2026. Evaluation data in, published member-club report out, with no manual step between: 250+ hours a year removed.

  • 10.4kBrightpearl contacts migrated into HubSpot and kept in sync

    Aid Through Trade · 2025. Brightpearl and HubSpot kept in step, with 10,399 contacts migrated.

  • 5 → 1dashboards replaced by one chat window

    Innovative Roofing Pros · 2026. MCP-connected agents that replace five dashboards with one chat window for a non-technical owner.

  • 4+plan types extracted automatically, from medical to life

    PlanYear · 2026. Playwright extraction across plan types, output as normalized datasets.

  • 2M+pages of discovery ingested for a single defense team

    CrossLogic Intelligence · 2026. An 83,000-page case fully extracted at about $0.003 per page, with cost metered per case and stage.

Which back-office work is worth automating

The work worth automating is the work your team repeats every week, where there is a clear workflow return. Invoicing, reporting, document handling and data extraction for finance and operations teams are the clearest examples in the case studies above.

It is not a fit when no process exists yet, when scope is undefined and nobody can make decisions, when an off-the-shelf tool already solves the problem cleanly, or when the goal is AI for marketing hype only. For the choice between a no-code connector and a custom build, see the Zapier or Make comparison.

Weighing a no-code connector against a custom build? Zapier or Make vs a custom integration: when each fits.

How does an engagement work?

  1. 01

    Discovery

    A 15-minute scoping call with direct questions about the goal and the constraints, so the written quote that follows matches what you need.

  2. 02

    Written quote

    Fixed scope, timeline, and deliverables on paper before any code is written.

  3. 03

    Build phase

    Daily async updates and weekly demo so you always see exactly where it stands.

  4. 04

    Launch + support

    Production deploy, documentation, and a 14-day fix window after handover.

Common questions

  • Who builds custom AI agents for back-office work?

    Your Private Dev is a one-person senior engineering service that builds AI agents, workflow automation, integrations and internal tools for operations and finance teams. The person you talk to does the building, on a fixed-scope written quote.

  • What can a custom AI agent do for an operations or finance team?

    Kintec's agent turns contractor invoice requests that arrive by email into finalized Business Central invoices. A roofing contractor's agents let the owner schedule, dispatch, invoice and look up records through one chat window. Both carry the task through rather than only answering questions.

  • What back-office work should be automated first?

    Work your team repeats weekly, where the workflow return is clear. At Kintec that was 70+ hours a month of contractor invoicing, and at Distinguished Clubs 250+ hours a year of report production.

  • Is a custom build better than Zapier or Make?

    Not always. A no-code connector is the better choice for simple, low-stakes steps between supported apps. A custom build fits two-way syncs, rollbacks, ERP depth and systems with no API. The comparison page sets out the criteria with examples.

  • How does pricing work?

    Pricing is fixed and agreed in writing before any code is written. After a 15-minute scoping call you get a written quote with fixed scope, timeline and deliverables within 48 hours. A fixed price means the cost is set by that written scope rather than by hours worked, so you know what you are committing to before the build starts. No price list is published, because every project is scoped on its own.

  • What is not a good fit?

    Four kinds of project are not a good fit: those with undefined scope and no decision-maker, processes that do not exist yet, cases where an off-the-shelf tool already solves the problem cleanly, and AI for marketing hype only. The service works best where a team repeats the same tasks weekly and there is clear workflow ROI. If you are unsure which side your project falls on, the 15-minute scoping call is the place to find out.

Have a similar problem? Let’s scope it.

Book a 15-min call Browse all work