GrowholdHow to use Growhold
Setting up your workspace, reading market signals, working on research, running agents, preparing product design and building knowledge.
Getting started
Growhold connects market intelligence, product strategy, planning and design work into one flow. Insights shows where demand is forming. Chat helps you think through what it means. Agents turn that thinking into delivered work. Knowledge compounds. Every step starts sharper than the last.
Create a workspace
Sign up and Growhold creates your workspace: a shared environment where market intelligence, conversation history, agent deliverables, and knowledge live together. During onboarding, Growhold can read your company's website and create initial context. You can also adapt this later by building out specialists and uploading your own materials.
Start with Insights
Insights shows live buying activity across markets: what people are paying to have solved, where demand is growing, and where supply is thin. It works whether you are entering a new market, developing a new product, or planning features for an existing one. The data updates continuously, so you can use it to validate a direction, spot an opportunity, or understand what buyers in a specific vertical are actively looking for.
Take it to Chat
Bring what you found in Insights into a conversation and explore what it means for your product or next feature. Chat draws from market data, your uploaded knowledge, and everything your workspace has already produced. You can explore a direction broadly, test a hypothesis, or work toward a specific output. When the conversation produces something worth keeping, ask Chat to create a card. That card flows into your workspace and is available to every agent run that follows.
Run an agent to go deeper
When you have a clear enough direction, an agent takes you to a finished deliverable. Open an agent from the store or your own library and start a run. Agents draw on market intelligence, workspace knowledge, and live web research to work through each step. Each step produces a card. The full set is the deliverable: a competitive analysis, a feature brief, a positioning document, a content plan. Each run builds on the context your workspace already holds.
Knowledge compounds
Every file you upload, every conversation you have, every run you complete adds to what your workspace knows. The next chat draws from it. The next agent run builds on it. You decide which cards and files become part of that knowledge: not every deliverable deserves to carry forward, and sometimes a direction needs to be rethought rather than compounded. Growhold gives you the controls to accept, exclude, and organise what the workspace learns from.
Insights
Insights gives you a live view of market demand: what buyers are paying to have solved, where demand is growing, and where gaps exist. Browse demand insights by vertical and buyer type. Useful when entering a new market, validating a product direction, or researching where to take an existing product next.
Insights overview
Insights maps live buying activity across markets, organized into demand insights by vertical and buyer type. Each demand insight represents a recurring pattern: a type of buyer, in a specific industry, looking for a specific kind of solution. Switch between demand areas to see demand through the lens of building products, automating workflows, or delivering services. Filter by vertical to narrow to the markets relevant to your work.
Reading the signals
Each demand insight represents a pattern confirmed across real purchasing signals. Its strength reflects how many signals match that pattern, how consistent the buyer language is, and how active the category has been. A strong demand insight indicates confirmed, active demand. A weaker one may indicate an early trend or a narrow niche. The buyer intent tags inside each demand insight show the language buyers use when describing what they need.
Demand areas, verticals, and demand insights
Insights organizes demand across three areas: build products, automate workflows, and deliver services. Each area reflects a different kind of buyer and a different kind of outcome they are paying for. Within each demand area, demand insights group by vertical so you can track the markets relevant to your work. Buyer intent tags inside each demand insight show the language buyers use when describing what they need, useful when forming a position or writing to a specific audience.
Opening a demand insight in Chat
Any demand insight in Insights can be taken directly into Chat. Click one and Growhold opens a conversation with that demand signal as context. From there you can explore what the pattern means for your product or feature roadmap, who is buying, and how your direction maps against it. Chat draws from the market data alongside everything already in your workspace.
How agents connect to Insights
Insights surfaces demand signals, but Chat and Agents both connect directly to the full demand database to dig deeper. A demand insight is the starting point: open it in Chat to explore what it means for your product direction, your positioning, or a specific buyer segment. Iterate in conversation until the picture is clear. When you are ready to produce a deliverable, run an agent: it draws on the same demand database alongside everything in your workspace, transforming your research into structured output. A competitive analysis agent can pull live demand patterns directly. A positioning agent can ground every sentence in the exact buyer language the database surfaces.
Chat
Chat knows your business through your knowledge base and delivered work. Research, plan, brainstorm, and create deliverables directly from the conversation.
How Chat works
Chat draws from everything in your workspace: uploaded files, delivered cards from past runs, and specialist context. Use it for research, competitive analysis, feature planning, brainstorming, or any question where your business context matters. Agents follow a defined path toward a specific deliverable. Chat is open-ended and can produce structured cards when you need a reusable output.
Research, plan, and ideate
Ask anything relevant to your product or business: explore a market angle, compare competitive positioning, plan a new feature, or brainstorm campaign concepts. Chat draws from uploaded files, past deliverables, and specialist knowledge. The more context your workspace holds, the more relevant the responses become.
Creating cards from Chat
Chat produces two kinds of output. For a quick deliverable, ask for a card directly: Chat creates a single structured card while your conversation continues in a split-panel view, with full versioning, editing, and export. Cards can include AI-generated images: ask for a moodboard, a UI mockup, or any visual and it generates inline alongside the text. For larger, more systematic work, Chat can also launch an agent from within the conversation. Agents follow a defined process and deliver a complete set of pre-structured cards across multiple steps, one card per step. Chat cards are ad-hoc and immediate. Agent cards are systematic and repeatable. Either way, the output flows into your workspace and is available to future conversations and runs.
Chat sessions
Keep different conversations organized with separate sessions. Each one maintains its own history, so you can have a running thread for product strategy, another for brand exploration, and another for competitive research. Filter by specialist to anchor a session in a specific domain.
Scoping context
Choose exactly what knowledge Chat draws from in each conversation. Select specific agents, specialists, or tags to narrow the focus. When your workspace covers multiple projects or product lines, scoping keeps answers grounded in the right slice of context. When a card panel is open, the scope button moves into the top bar so it stays accessible.
Getting the most out of Chat
Start broad, then narrow. Ask Chat to explore a market, then scope to a specific agent's knowledge and dig deeper. Attach specialists to bring domain expertise into the conversation: a brand specialist keeps answers on-brand, a research specialist adds depth. Ask for a card when you want a reusable deliverable instead of a chat reply. Use sessions to keep separate threads for different projects. And remember: every card you create and every file you upload makes Chat smarter for next time.
Knowledge
Everything Growhold knows about your business lives in two places: the Knowledge Base holds what your team brings in, and Work Results holds what agents deliver. Both feed every Chat conversation and every agent run.
Knowledge Base and Work Results
The Knowledge page has two tabs. Knowledge Base holds everything your team brings in: uploaded files, pasted notes, forwarded emails, and scraped pages. Work Results holds the cards agents have delivered that you have added to your knowledge. Both are active context: every Chat conversation and every agent run draws from both. Upload a competitor report and agents can reference it immediately. Accept the cards from a finished run and the next run builds on them. The more your workspace accumulates, the sharper every output gets.
Uploading files
Drag and drop files or click to browse. Text, PDF, Word, markdown, and other common formats are supported. You can also paste text directly. Product specs, brand guidelines, market research, strategy docs, competitor pages, meeting notes: anything that helps Growhold understand your world.
Applying knowledge
Once uploaded, apply files to specific specialists or agents. This tells Growhold which context to use during runs. The same file can serve multiple specialists. A brand voice document might inform both your content and design specialists.
Work Results
The Work Results tab shows the agent-delivered cards that are part of your workspace knowledge. Cards are grouped by the run that produced them so you can see which agent generated them and navigate back to the original work. You control what enters knowledge: when a run finishes, you can accept its cards into knowledge or keep them separate. Remove individual cards or a full run from knowledge when they are no longer relevant. Everything in Work Results is searchable by keyword or semantic query, and available to every future Chat conversation and agent run.
Email and webhook import
Keep your knowledge fresh without manual work. Forward newsletters, research reports, customer feedback, or any recurring content directly into your workspace via email or webhook. New knowledge flows in automatically.
Searching knowledge
Find anything in your knowledge using natural language. Growhold uses semantic search, so you will find relevant content even when the exact words do not match. Search works across both Knowledge Base files and Work Results cards in their respective tabs.
Agents
You have a goal: a competitor analysis, a feature brief, a content plan. Agents work through it step by step and deliver the result as structured cards.
What is an agent
An agent is a repeatable path to a specific deliverable. You define the goal — a competitive analysis, a feature brief, a content calendar — and the agent works through it step by step. Each step produces one card, and the full set is the deliverable. Agents draw from your knowledge base and specialists, so the output reflects your specific context. Every card flows back into your workspace, available to future runs and conversations.
Create your first agent
Go to Agents and click Create. Describe what you want to accomplish and Growhold builds the agent for you, complete with name, instructions, and steps. Want something ready-made? Browse our agents and install one with a single click. Prefer full control? Build it manually from scratch.
Run your first agent
Open an agent and hit Start run. In conversational mode, you work through each step together, giving input, adjusting direction, and reviewing each card as it takes shape. In autonomous mode, the agent handles everything on its own. Either way, each step delivers a card, and the complete set becomes part of your workspace knowledge, ready for the next run or conversation.
Agent steps
Every agent is made up of steps. Each step has its own instructions, inputs, card template, and suggested cues, and it produces exactly one card. During a run, the agent moves through steps in sequence, each one building on what came before. You can guide each step directly, provide additional context, or let it work autonomously. The card it delivers can be edited, versioned, and refined after delivery.
Input fields
Some steps need specific information upfront, like a target audience, a competitor name, or a budget range. Input fields let you define exactly what each step needs: text fields, dropdowns, numbers, checkboxes. Dropdowns can even generate options dynamically based on your workspace context. It's structured input without the rigidity.
Card configuration
Control what each step's card looks like by defining a template in markdown. Include text sections, tables, lists, statistics, and image placeholders. Use {{image: description}} to generate visuals inline: UI mockups, brand assets, concept illustrations, moodboards, diagrams, or any image the description calls for. The template shapes the card, aiming for concise, visual, easy-to-scan output rather than long-form documents. Every run produces a card that's immediately useful.
Agent settings
Name your agent, describe its purpose, set the deliverable format, and write instructions that shape its focus, tone, and boundaries. Add tags to keep things organized and set a cover image to make it recognizable at a glance.
Step settings
Fine-tune each step independently: its name, description, and the instructions that guide its behavior. Define what it focuses on, the format it follows, and how it responds. Each step can be adjusted without affecting the rest of the agent.
Agent improvements
Growhold watches how your agents perform and suggests improvements: sharper instructions, better card formats, additional steps. Review each suggestion, apply it with one click, or skip it. Every change is tracked so you always know what was updated and why.
Connections
Give your agents access to live data. Enable web search, page scraping, news feeds, social media, or video platforms and toggle each one on or off per agent. When a connection is active, the agent can pull fresh information during runs instead of relying only on what's already in your workspace.
Work overview
See all your agent runs in one place. Check what's in progress, what's done, and where things stand. Each run shows its status, which agent it belongs to, and how far it's progressed through its steps. Filter by agent, status, or tag to find exactly what you're looking for.
Agents Store
Don't want to build from scratch? The agents store has ready-made agents for common workflows like competitor analysis, SEO audits, content calendars, and product briefs. Search by name, filter by category, and install any agent into your workspace with one click. They arrive fully configured. Tweak them to fit your context or run them as-is.
Specialists
Agents follow a process. Specialists carry the domain knowledge that shapes how the agent applies it. Attach specialists to runs to ground every deliverable in your specific context.
What is a specialist
Specialists give your agents domain knowledge. A design specialist might carry your brand palette, image guidelines, and voice. A product specialist might hold your roadmap, positioning docs, and competitive landscape. When you assign specialists to a run, every step draws on that context. The output reflects your specific product, market, and strategy.
Creating specialists
Describe the domain you need and Growhold generates a specialist profile. You can also import one from a URL: Growhold reads the page and builds a specialist from its content. Or build one manually for full control. During onboarding, Growhold can create initial specialists from your company's website to get you started.
Configuring specialists
Give your specialist the knowledge it needs: brand guidelines, product specs, sample content, color palettes. Edit its name, instructions, tags, and image. When this specialist joins a run, everything attached to it shapes how the agent responds. A brand voice doc attached to a content specialist means every piece of writing stays on-brand.
Guide: process and brand instructions
The guide is where you write down how your specialist should think. Add instructions for brand voice, naming conventions, content structure, taxonomy rules, review processes, or any logic your team follows. Think of it as the specialist's internal playbook. When an agent uses this specialist, it follows these instructions just like a new team member would follow your onboarding docs.
Knowledge: documents and files
Attach documents your specialist needs to do its job well. Strategy decks, product specs, meeting notes, customer research, email threads, competitive teardowns. Any file that carries institutional knowledge belongs here. The more relevant material you provide, the more grounded and accurate the specialist's contributions become during agent runs.
Visuals: design assets and palettes
Upload color palettes, brand sample images, design system references, or visual guidelines. When your specialist is involved in generating images or visual content, these assets inform the output. A product marketing specialist with your brand colors and style references produces visuals that match your identity instead of generic stock imagery.
Cards
Every step produces a card: a focused, visual piece of output you can review, edit, and build on. An agent's full deliverable is the collection of cards from all its steps.
What are cards
A card is the output of a single step. When an agent runs, each step produces one card: a competitor analysis, a mood board, a content calendar, a set of recommendations. Together, those cards form the agent's complete deliverable. Cards are structured for quick scanning: concise insights, visual layouts, key statistics, and defined sections. Once delivered, each card flows back into your knowledge base and is available to Chat and future runs.
Version history and editing
Cards aren't frozen after delivery. Edit any card with AI assistance or make manual changes directly. Every edit creates a new version, so you can always see how a card evolved from first draft through revisions. Regenerate a card with updated context, refine the wording, add details, or restructure the layout. The version history gives you a full trail of changes and the confidence to iterate freely.
Visual content: images in cards
Cards can include AI-generated images alongside text, tables, and charts. Add image placeholders to any card template using {{image: description}} and the image generates inline when the card is produced. You can generate UI wireframe mockups, brand moodboards, product concept illustrations, social media visuals, content assets, and more. Configure style, mood, subject, setting, and perspective in the step settings to keep visuals consistent across runs. If a specialist carries your brand colors and sample images, generated visuals stay aligned with your visual identity automatically.
AI image editing
Every generated image in a card can be edited with AI. Hover over an image to reveal the edit button, then describe the change: adjust the color palette, swap an element, refine the composition, or apply a new style. Paint a mask over the part you want to focus the edit on, then describe what should change there. Each edit produces a new version in the history strip so you can compare results and step back at any point. Control output quality, aspect ratio, background transparency, and how closely the result should preserve the original image. Image editing works on any card, whether it came from Chat or an agent run.
Charts and data visualization
Add interactive charts to any card: pie, bar, line, area, and heatmap. Define chart placeholders in the card template using {{chart:slug}} and configure each chart's type, axis labels, and seed data in the step settings. During a run, the step populates charts with real data from your workspace context. Charts render inline and update with each card version, turning raw numbers into clear visual insights.
Exporting and using cards
Review delivered cards from the run page, grouped by step. Cards render with full formatting and can be exported for use outside Growhold as reports, briefs, plans, creative assets, or whatever your workflow needs.
Starred cards
Star your most important cards to keep them within reach. Click the star icon on any card's toolbar to mark it. Starred cards appear in a dedicated grid view on your dashboard, accessible via the Starred button at the top. The grid shows each card as a visual tile with the agent name, a content preview, and a timestamp. Click any tile to jump to the full card inside its run. Use starring to bookmark your best deliverables, track key outputs across multiple agents, or curate a personal collection of reference material.
Schedule
Set your agents to run on a schedule. Define the frequency, pick the specialists, and Growhold handles the rest. Your workspace keeps building knowledge while your team focuses on other work.
How scheduling works
Create a schedule for any agent and it runs automatically at the times you set. Choose from preset frequencies like daily, weekdays, weekly, or twice a week, or write a custom cron expression. Each scheduled run uses the agent's current configuration, specialists, and knowledge, so output stays relevant as your workspace evolves.
Creating a schedule
Go to Schedule and click Create. Pick an agent, name the schedule, choose a frequency, and select which specialists to include. Add optional instructions to focus the run on a specific angle or topic. Enable or disable any schedule at any time without deleting it.
Managing schedules
See all your scheduled runs in one place with next run time, run count, and status. Pause a schedule when you need a break, resume it later, or delete it when it's no longer needed. The dashboard sidebar also shows your upcoming scheduled runs so you always know what's next.
Dashboard
Your workspace at a glance. See what's happening, what's coming up, and jump straight into a conversation.
Dashboard overview
The dashboard shows your recent activity feed, upcoming scheduled runs, and knowledge queue in one view. See which agents have delivered, what's in progress, and what's queued for processing. The activity card tracks your credit usage over time with a visual heatmap.
Ask box
Start a conversation directly from the dashboard. Type a question or attach a file, and Growhold opens a new chat session scoped to your workspace knowledge. No need to navigate to Chat first.
Activity feed
Your feed shows everything that happened in your workspace: delivered agent runs, active runs, chat conversations, and newly uploaded knowledge. Filter by tag or search to find specific activity. Each item links directly to the full run, conversation, or file.
Notifications
The bell icon in the top corner keeps you informed without interrupting your work. When an agent run completes or fails, a notification appears with an unread badge. Open the panel to see recent activity at a glance: green for completed runs, red for failures, blue for new cards. Click any notification to jump straight to that run. Notifications mark themselves as read when you open the panel, or you can clear them all at once.
Plans and billing
Simple pricing, full access on every plan. Pick the usage level that fits your team.
Available plans
Three plans: Starter ($49/mo, 15,000 credits), Growth ($99/mo, 30,000 credits), and Scale ($199/mo, 60,000 credits). Every plan includes unlimited users and full access to all features, agents, and the agent store with no feature gating. Start with a 7-day free trial on any plan.
Usage tracking
Your current credit balance and usage are visible directly on the plans page. You can always see how many credits you've used and how many remain in your billing cycle.
Managing your subscription
Upgrade, downgrade, or update payment methods through the billing portal. Check your current plan, usage against limits, and renewal date, all in one place.
Settings
Set up your workspace and manage your account.
Workspace settings
Name your workspace, set an image, and make it yours. The workspace image shows up in the sidebar so your team can identify the environment at a glance. Upload a custom image or generate one from a color palette.
Account settings
Update your profile name and change your password. Your email address is set during sign-up and shown as read-only.
Selecting US or EU AI providers
Workspace admins can choose which AI infrastructure region the entire workspace uses. Open Workspace Settings and find the US and EU AI Models toggle. Select United States to use Anthropic and OpenAI, or European Union to route all AI requests through Mistral and Blackforest Labs hosted within the EU. The setting is workspace-wide and takes effect immediately for all users. Members can see the active region in the top-right corner of the app but cannot change it.
API & Developers
Connect Growhold to external tools, AI assistants, and automated workflows through the REST API and MCP server.
API keys
Create and manage API keys from the API & Devs page in workspace settings. Each key starts with gh_ and can be scoped with permissions for specific domains: knowledge, runs, schedules, agents, and specialists. Keys are hashed before storage, so copy your secret immediately after creation.
REST API
All API calls use a single POST endpoint with an action-based routing pattern. Send JSON with an action field like agents.list or runs.start, and include your API key as a Bearer token. Detailed endpoint documentation is available in the API Reference.
MCP Server
The MCP server lets you connect Growhold directly to Claude Desktop, Cursor, or any MCP-compatible AI client. It exposes all API actions as tools using the Model Context Protocol (JSON-RPC 2.0). Add the MCP server URL and your API key to your client configuration, and your AI assistant can manage knowledge, start agent runs, and control schedules.