Your AI API, your rules: Introducing Custom AI Job instructions

The problem with one-size-fits-all AI

SharpAPI's endpoints are built to work out of the box. You send a resume, you get a score. You send a job title, you get a description. That's the promise, and it holds for most use cases.

The workaround until now was wrapping our API in your own middleware, manually injecting context before every call. That works, but it's boilerplate you shouldn't have to write.

We shipped a native per-account, per-endpoint instruction system. You write the context once. We inject it into every relevant AI request automatically, at the right layer, without touching your API calls.

How it works

Every SharpAPI account now has a Customize AI Jobs section in the dashboard. It lists all available AI job types grouped by category, from HR and recruitment endpoints to e-commerce and content tools.

For each job type, you can write a free-form instruction. Think of it as a persistent system note to the underlying model: "for every resume scoring request I make, apply these priorities." Once saved and toggled active, that instruction is automatically injected before the AI processes your request.

Step 1: Open the dashboard and go to Customize AI Jobs Find it in the sidebar, right below Custom Workflows. All job types are listed and grouped by category.

Step 2: Expand the endpoint you want to customize Each card has an inline editor. Write your instruction in plain language, no special syntax required.

Step 3: Save and activate You'll see a confirmation prompt before saving, since changing instructions affects your output. Toggle it on, and you're done.

Step 4: Your API calls stay exactly the same No changes to request format, headers, or parameters. The customization happens server-side, invisibly, for every matching request.

What this looks like in practice

Say you're using the Resume Scoring endpoint to screen candidates for a Cloud Architect role. Without a custom prompt, the model scores resumes against a generic template. With one, you can shift those priorities permanently for your account:

Weight cloud certifications (AWS, GCP, Azure) significantly higher than general programming experience. Penalize resumes with no evidence of distributed systems work. Flag any candidate with fewer than 3 years of hands-on infrastructure experience regardless of their total years in software.

Your API call stays unchanged. But behind the scenes, your custom instruction is already baked in. Every candidate's resume is scored against your actual hiring criteria, not a generic baseline.

What you can customize

Tone and style Lock content generation to your brand voice. Formal, conversational, regional, industry-specific -- it's up to you.

Scoring weights Boost or suppress specific signals in scoring endpoints. Relevant for resume scoring, content quality checks, and more.

Output structure Tell the model to always include or exclude certain fields, use specific phrasing, or follow your internal format.

Domain context Give the model background it otherwise wouldn't have: your industry, your audience, your product category.

Note on Custom Workflows: Custom Workflows already support user-defined prompts as a first-class feature. Custom Job Prompts do not apply to Workflow requests -- they're scoped to predefined endpoints only.

Managing your prompts

Action What it does

Save a prompt Stores the instruction and activates it immediately

Toggle off Disables the prompt without deleting it. Default AI behavior restored.

Toggle on Re-activates a previously saved prompt

Delete Removes the prompt entirely. Endpoint returns to default behavior.

Every save triggers a confirmation dialog. This is intentional: custom prompts can meaningfully change your outputs, and we want the decision to feel deliberate rather than accidental.

Heads up: Custom prompts are powerful, but they can also produce unexpected results if they conflict with the underlying endpoint logic. If your results start looking off, toggling the prompt off is the fastest way to confirm whether it's the source of the issue.

What this means for teams and integrators

If you're building a product on top of SharpAPI and serving multiple clients, custom prompts let you pre-configure the AI layer for each account without maintaining separate middleware or prompt injection logic on your end. Each account operates independently: one customer's scoring weights don't bleed into another's.

For solo developers, this is simply a way to stop re-writing the same context in every API wrapper you build. Write it once, use it forever, change it whenever your needs shift.

Get started

Custom Job Prompts are available to all accounts with API access. No plan upgrade required. Log into your dashboard, find Customize AI Jobs in the sidebar, and start tuning.

Open Dashboard