Performance monitoring · Workflow board
Creative fatigue watch
Every morning, find ads with rising frequency and falling CTR, judge fatigue with AI, and propose pausing the worst, with your approval.
Ready-made agents, connectors for your ad accounts, and the models and skills Hawky builds with.
Official agents, built for Meta and Google ads
Performance monitoring · Workflow board
Every morning, find ads with rising frequency and falling CTR, judge fatigue with AI, and propose pausing the worst, with your approval.
Creative · Live board
This week's numbers against last week, the creatives that deliver best once they have real spend, and where every creative sits on spend against efficiency.
Budget · Live board
Every creative with real spend on one matrix against the account's own efficiency: what to give more budget, what is wasting it, and which ads are tiring.
Creative · Live board
Every creative with your brand's own columns and naming filters, written for you by the agent. Ask it to change anything.
Analysis · Live board
Your creatives grouped by an AI creative tag you pick: what each value delivers, its share of spend, and its best creative.
Reporting · Live board
Spend, results and efficiency against the previous period, the daily trend, where spend goes by objective and every campaign side by side.
Creative · Live board
Which creatives spend and deliver: totals, the top creatives by spend, and the daily trend.
Competitors · Live board
Your tracked competitors' ads, live: who is most active, what they push now, their proven ads, and the hooks and formats they lean on.
Reporting · Workflow board
Yesterday's spend, results and cost per result against your 7-day average, as a board you open with your coffee.
Built by marketers in their own workspace, then published to the Hub
Performance monitoring · Workflow board
Last week against the week before for each Meta account, with the ROAS change broken down into CPM, CTR, CVR and AOV, campaign drivers and an AI-written cause.
Performance monitoring · Workflow board
Judges every active Meta ad on the result its ad set is optimised for, 7-day against 28-day, and lists ads to pause, ads dipping and ads to push.
Competitors · Workflow board
Tracks creator collab ads from competitors over the last 90 days: creators per brand, live activity, creators shared across brands, new creators, and a full directory.
What Hawky follows when it designs a page, draws charts or writes copy
Design skill
One story across swipeable slides: the story first, one theme from the brand, one background that runs across every slide, then each slide's words and logo.
Charts skill
The right chart form for the question, the finding in the title, colour that encodes, and the usual mistakes to avoid.
Design skill
Intentional visual design for a page you write (layout, type, colour, copy), so it does not read as a templated default.
Analysis skill
Turns what a chat found into a client-ready Hawky report: a short version up front, then the evidence, the fixes and the ask, in Hawky's house style, as HTML and PDF.
Where the money goes. Spend, results and creatives from every channel you buy on.
No API on these platforms, so Hawky reads the report you upload or drop on SFTP.
Where results reach a person: a channel, a phone, an inbox, or your own system.
App installs and in-app events by media source, from your MMP.
Leads, deals and stages: the outcome a lead-gen ad is really judged on.
Sessions, events and conversions on your site and app.
Orders, revenue, subscriptions and owned-channel revenue.
Your own tables, plans and files, including systems with no API at all.
Sign in once and Hawky's agents can use the tool, while they build and in every run.
Writing, analysis and the reasoning behind every agent run
Long-form reasoning and writing: briefs, summaries and the cause behind a number.
General-purpose writing and analysis across agents and chat.
Analysis over large exports and long account histories.
Long-context model for agents and research.
Open-weight multilingual family, from small models to frontier ones.
Open-weight reasoning and coding model.
Open-weight model built for agentic tasks and tool use.
Long-context model for agents and coding.
Static creatives and edits, in your brand
Generate and edit images from plain instructions, keeping the product or person the same across edits.
Follows detailed prompts closely and renders legible text inside the image.
High-resolution image generation and editing.
Photoreal generation and in-context edits from a reference image.
Open-weight image model, strong at rendering text, including Chinese.
Strong at type inside images, for offer-led statics and banners.
Video creatives, from a prompt, a product shot or a winning static
Google's video model: text and image to video, with sound.
OpenAI's video model: text and image to video, with sound.
Text and image to video with strong, natural motion.
Text and image to video, including multi-shot sequences.
Image to video with control over camera and motion.
Text and image to video for short, punchy clips.
Open video model for text and image to video.
Animates a person from one photo and an audio track: talking avatars for UGC-style ads.
Search the Hub before you build. Most jobs a performance team repeats every week are already here: fatigue, pacing, lead quality, competitor watch.
Hawky rebuilds the agent for your brand, asks what only you know, like your targets and naming, and tests it on your own accounts.
Change anything by asking in chat. The workflow behind it and the board in front of it are yours to edit, and nothing acts without your yes.
When you build something good, publish it to the Hub from the agent's menu, so the next marketer doesn't start from zero.
The Hub is the public library inside Hawky. It holds ready-made agents, skills, connectors and models, from Hawky and from the marketers who use it. Your own running agents live in Agents; the Hub is where you find new ones.
Hawky rebuilds it for your brand. It connects to your accounts, asks what only you know (targets, naming, who approves), and tests it before it goes live. You get your own copy, with a board your team opens and a workflow underneath that runs on a schedule.
Only with your approval. Agents that propose changes, such as pausing an ad or moving budget, wait for someone to say yes before anything changes. The rest only read and report.
Hawky publishes the official agents and skills. Anyone using Hawky can publish an agent or a skill they built, so the library grows with the community.
Pick the model for the job. For writing and analysis, the Hub has the frontier language models from the US and China: Claude, GPT, Gemini, Kimi, Qwen, DeepSeek, GLM and MiniMax. For creatives, it brings the popular image models (Nano Banana, GPT Image, Seedream, FLUX, Qwen-Image, Ideogram) and video models (Veo, Sora, Kling, Seedance, Runway, Hailuo, Wan, OmniHuman). New models are added as they ship.
Add a ready-made Hawky agent, connect your accounts, and it runs every day. See it on your own accounts in a live demo.