What Amplifiers can do
Amplifiers exposes one small, stable set of tools to your assistant. Those tools are the doorway; the catalogue behind them — hundreds of prompts and data integrations — changes constantly without your connector needing to change.
That design is deliberate. Your assistant never has to hold a list of hundreds of tool definitions. It calls find_amplifiers with a description of what you want, and gets back the handful that fit.
The tools the connector exposes
Discovery
| Tool | What it does |
|---|---|
find_amplifiers | The single entry point. Takes a natural-language description of what you're trying to do and returns matching prompts and tools, ranked. Everything starts here. |
setup_amplifiers | A personalised first-run walkthrough — what's available on your plan and what to try first. |
amplifiers_dashboard | What's most used across the platform and by you, in an interactive panel. |
Running an amplifier
| Tool | What it does |
|---|---|
use_prompt | Fetches a prompt template and its interview instructions. Your assistant asks you for the details the template needs, then runs it. |
run_tool | Runs a catalogue data tool — the YouTube, LinkedIn, Reddit, search, and company-lookup integrations, and everything else API-backed. |
save_prompt / draft_save_prompt | Saves a prompt of your own to your personal library so it turns up in future searches. |
Images
| Tool | What it does |
|---|---|
generate_image | Generates an image from a prompt, and renders it in an interactive panel inside the conversation. |
generate_image_batch | Generates several variations in one go. |
refine_image | Edits an image already in the session — "make the background warmer", "remove the text". |
show_style_gallery | Shows a visual gallery of styles for an image prompt so you can pick by eye rather than by description. |
upload_reference | Opens a panel for uploading your own reference material — a logo, a photo, a product shot — for the model to work from. |
get_image_session | Reopens an earlier image session so you can keep refining days later. |
Long-running work
Some tools take longer than a chat turn allows. Those return a job you can wait on.
| Tool | What it does |
|---|---|
wait_for_job / get_job_status | Waits for, or checks, a long-running job. |
cancel_job | Stops one. |
Account
| Tool | What it does |
|---|---|
account_status | Your plan, your credit balance, and a link to manage your subscription. |
Collaborative review
| Tool | What it does |
|---|---|
get_feedback / wait_for_feedback / close_feedback_session | The Review Room: share a link, collect comments from other people on a batch of drafts, and bring their feedback back into the conversation. |
The catalogue behind them
The catalogue is where the substance lives, and it grows every week.
Prompts — expert-crafted templates for LinkedIn posts, cold emails, landing-page copy, résumés, product descriptions, code review, image generation and more. They're engineered to produce noticeably better results than writing a prompt from scratch, and the assistant interviews you for the specifics before running one.
Data tools — API-backed integrations that fetch real information:
- Video and audio — YouTube transcripts, video and channel metadata
- Social — LinkedIn profiles and posts, Reddit threads and comments, X/Twitter, Instagram, Facebook Ad Library
- Web — search, grounded web research, article extraction
- Business — company lookup and enrichment, domain and social-handle availability
- Generic LLM access — routing a sub-task to a specific model
Image workflows — style galleries, reference-image upload, batch generation, and refinement that persists across sessions and devices.
Browse everything at getamplifiers.com/catalog.
What Amplifiers does not do
Being explicit, because it matters for how you evaluate the connector:
- It does not read your conversation. Only the specific call your assistant makes reaches us.
- It does not access your device, your files, your email or your calendar. The only files we ever see are ones you deliberately upload.
- It does not act on its own. Every call is initiated by your assistant in response to something you asked for. There are no background agents and no scheduled actions.
- It does not send anything on your behalf — no emails, no posts, no messages.
- It does not train AI models on your data.
See Data and permissions for the full picture.