# AI implementation and custom AI development, one working piece at a time

> AI implementation services and custom AI development for small business: one AI feature, agent or integration at a time, tested on your examples.

Canonical: https://app369.com/service/ai-implementation/

For clinics, law firms, agencies, SaaS teams and real estate offices: Simon Dziak builds AI for the task that eats the most hours, one fixed-scope piece at a time. Each AI feature or agent starts with a one-page spec you approve and is tested on your own examples before launch; every piece is delivered as code in accounts you own.

A feature or agent that reads, sorts, drafts or answers is the AI Feature or Agent Build, a fixed $2,500 delivered 10 business days after the spec is agreed and access arrives. Connecting two of your systems is the API Integration, a fixed $1,200. For a steady list of AI work on apps and tools you already have, the Monthly Dev Subscription Studio plan is $3,995 a month, one request at a time, delivered as pull requests.

Fixed prices

[**All fixed-price offers**](https://app369.com/fixed-price/)

- [$2,500**AI Feature or Agent Build**One LLM feature or agent wired into your app or tools, with an eval set that shows how well it works.**See scope and start**](https://app369.com/fixed-price/ai-feature-development/)
- [$1,200**API Integration**Two systems connected through their APIs, with retries, logs, failure alerts and tests.**See scope and start**](https://app369.com/fixed-price/api-integration/)

Something else? [**Describe your project**](https://app369.com/contact/)

Where AI goes to work

## **Pick the first piece, then the next**

- 01 ### **Answering leads** Calls, website chat or web forms answered on one channel, the lead qualified and booked, and a person brought in with a summary. [**AI receptionist and answering service**](https://app369.com/service/ai-receptionist/) [**Speed to lead and missed-call text back**](https://app369.com/service/speed-to-lead/)
- 02 ### **Automating the work your team repeats** Lead intake, document extraction, email triage, tool syncs and weekly reports, plain code where the steps are fixed and AI only where the input varies. [**AI automation services**](https://app369.com/service/ai-automation-services/)
- 03 ### **AI features inside your app** A clinic’s intake form summarizer, a law firm’s clause finder or a real estate office’s listing reply drafter, built into the web or mobile app you already run. [**AI app development and integration**](https://app369.com/service/ai-integration/)
- 04 ### **Agents with limited tools** An agent that looks up a case, a booking or an order through your API and prepares the next step for a person, limited to the actions in its spec, with write actions gated. [**Hire an AI agent developer**](https://app369.com/service/hire-ai-agent-developer/)
- 05 ### **Not sure which piece comes first** Write in with the task that eats the most time. Simon replies in writing with the piece that fits, its scope and its fixed price, before you pay anything. [**Describe your task**](https://app369.com/contact/)
- 06 ### **Serving Miami, Tampa and the US** Simon works from Miami and implements AI for businesses across Florida and the rest of the US, by email and shared documents. [**App369 in Miami**](https://app369.com/locations/miami-florida/) [**App369 in Tampa**](https://app369.com/locations/tampa-florida/)

How each piece is built

## **Spec, build, test, hand over**

Anthropic’s guide to agents recommends “finding the simplest solution possible, and only increasing complexity when needed”. Each piece follows that: the smallest build that does the job, measured before it goes live.

- **✓**### **A one-page spec first** The task, the inputs and outputs, the tools it may call and how success is measured, agreed in writing before any build starts.
- **✓**### **Tested on your examples** Each AI feature or agent gets an eval set of up to 50 cases from your own examples, with a scored run before handover, so you see how well it works before your customers do.
- **✓**### **Gated writes and full logs** For each AI feature or agent, tool access is limited to the actions you approve, write actions are gated, and prompts, tool calls and token cost are logged.
- **✓**### **Handed over in your accounts** Built on your own model provider account and delivered as a pull request with a README and handover notes, so you own the code and the keys.

- Claude API
- Gemini API
- MCP
- Webhooks
- Node.js
- TypeScript
- Firebase
- Google Cloud

FAQ

## **Questions before you start**

<details>

<summary>**What do AI implementation services cost? **</summary>

Each piece has a fixed price. An AI feature or agent is the AI Feature or Agent Build, $2,500, delivered 10 business days after the spec is agreed and access arrives. Connecting two systems is the API Integration, $1,200, delivered 5 business days after API docs and test credentials arrive. Ongoing work is the Monthly Dev Subscription Studio plan, $3,995 a month, sold on its own page rather than in the price box above. Model usage is billed to your own provider account.

</details>

<details>

<summary>**What does custom AI development mean here? **</summary>

Custom code around Claude or Gemini, not a model trained from scratch. It is software around an existing model: the Claude or Gemini API, or another model API you already use, wired to your data and tools with your rules. It does not include training or fine-tuning a model of your own.

</details>

<details>

<summary>**Where should a small business start? **</summary>

With one task your team repeats often, that follows a few clear rules, and where a mistake is cheap to catch. Answering leads, sorting incoming email and pulling fields out of documents are common first picks.

</details>

<details>

<summary>**Do I need an AI strategy or a consultant first? **</summary>

No. Describe the task by email and the scope is agreed in writing before you pay. If you want to talk it through, consultations are paid: $100 per 15 minutes, paid upfront.

</details>

<details>

<summary>**How do I know it works? **</summary>

Each feature or agent comes with an eval set of up to 50 cases from your examples and a scored run before handover. No model is right every time and no accuracy score is guaranteed, so a task where a mistake is costly keeps a person in the loop.

</details>

<details>

<summary>**Can a clinic or law firm use it? **</summary>

Yes, with care. A clinic is usually a covered entity under HIPAA, and whether each model and tool vendor will sign a business associate agreement depends on the vendor and plan. A law firm has its own client-confidentiality duties. The spec lists every vendor that would see client or patient data, and you and your counsel decide before launch. For a clinic or law firm the eval cases are made-up or de-identified unless you and your counsel decide otherwise. This is not legal advice.

</details>

<details>

<summary>**Who owns what gets built? **</summary>

You do. It runs on your own model provider account and is delivered as a pull request to your repository, with a README and handover notes. If you don’t have a repository or cloud account yet, say so when you write in; setting them up is quoted separately in writing.

</details>

<details>

<summary>**Who does the work? **</summary>

Simon Dziak, who founded App369 and built Workspace369’s integrations, background jobs and AI features, does every part of the work.

</details>

### **Sources**

- [Anthropic: Building effective agents (Dec 19, 2024)](https://www.anthropic.com/engineering/building-effective-agents)

Not a fixed-price job?

## **Describe it and get a written quote**

Send what you need built or fixed. You get a written scope, a fixed price and a timeline; Simon replies by email, usually within one business day.

[**Describe your project **→](https://app369.com/contact/) [**Instant quote**](https://app369.com/instant-quote/)

Prefer a call? Consultations are $100 per 15 minutes, paid upfront.