Answer a customer, update a guide, plan a fix, or follow up with a lead. Give agents the knowledge and tools for the job, then choose what runs automatically and what needs review.
Open source · Self-host free, or let us run it
Choose an agent for the job. It can use the relevant conversations, docs, and tasks already in Helpin, along with the tools you allow.
Answer from your docs and earlier conversations, check in after a reply, and hand over when a person is needed.
Turn a request or product idea into a clear scope and a plan your team can review.
Read the requirements and code, work out the changes, and prepare a task before coding starts.
Write the change, run the tests, and open a pull request in your connected GitHub or GitLab repository.
Check the proposed code for bugs, missing tests, and requirements that still need work.
Update guides from customer questions and product changes. Use browser screenshots and recordings where the steps need showing.
Find who needs attention, update customer records, and prepare a follow-up using the account’s conversations and open work.
Use what customers ask for to prepare launch messages, campaign ideas, and clearer product explanations.
Describe the work you want done, from reviewing code to following up with leads. Helpin AI helps you create the agent or flow, then you review its instructions, tools, and steps before it runs.
Ask about a customer or give Helpin a task. Helpin AI finds the relevant history, uses the tools you’ve allowed, and brings in specialist agents when the work needs them.
Ask Agent can have specialists check the code, customer report, and docs together. Their findings shape the plan before anyone starts changing things.
Bring the customer’s question, relevant guidance, and linked work into the same answer.
Your AI agents check the report, code, and guide. Use their findings to choose the next action.
Connect external tools through MCP (beta) and select what each agent may use. Access stays limited to the tools you grant.
Scribe plans the task. Forge writes code and tests. Lens reviews the change. Keep the original report close so everyone can check whether the fix solves it.
Coding agents need a connected GitHub or GitLab repository and Agent Runtime, on Helpin Cloud or self-hosted.
Explore coding agentsFollow the run, inspect the results, and see what needs a decision. Your team keeps a place to review past work, check the evidence, and pick up anything an agent could not finish.
Set the instructions, skills, and model for the work. Configure a specialist or create an agent for your own process.
Select its workspace and external tools. Decide which actions can proceed and which need approval.
Run an agent on a schedule or when a task changes stage or a pull request opens or merges. You can also start it yourself. Review the run and results in Helpin.
Eight specialists cover support, planning, coding task planning, coding, code review, docs, CRM, and marketing. Ask Agent is your starting point for finding answers and coordinating their work.
It is the assistant your team talks to inside Helpin. Ask a question or give it work. It can use permitted tools directly or bring in specialists when needed.
Yes. Ask Agent can coordinate specialists, run independent steps at the same time, and bring their findings back into the conversation.
Yes. Forge works in a connected GitHub or GitLab repository, runs the tests, and opens a pull request for your team to review. Coding agents need a connected repository and Agent Runtime, on Helpin Cloud or self-hosted.
Yes, in beta. Connect an external MCP server and select which tools an agent can use. The connection, tool permissions, and approval rules determine what it may do.
Yes. Configure the instructions, skills, model, tools, and approvals for a specific job.
Yes. Assign a task to an agent, or use an automation flow to start one when a task enters a stage, when a pull request is opened, merged, or needs review, or on a schedule.
Yes. Self-hosted Helpin is the complete product with every agent, coding agents included, free under AGPL-3.0 with no plan limits. You run the installation and connect your own AI provider.
Start with a useful task, give the agent the right tools, and review the result in the same workspace.
14-day free trial · No card required