Question 1

What are AI agents in Neuphlo?

Accepted Answer

AI agents are autonomous specialists that handle day-to-day project work: creating tasks, writing documentation, managing sprints, running tests, and more. They work using the same tools your team would, but operate around the clock. You review, override, and approve their work. Every action is logged in a tamper-evident audit trail.

Question 2

Are AI agents included on the free plan?

Accepted Answer

Yes. All plans include AI agents with the full audit trail, confidence scoring, and one-click rollback. The free Starter plan has the same agent capabilities as Plus. The limits are on tasks (50), articles (10), and members (1). Enterprise adds self-hosting, SSO, and compliance documentation.

Question 3

How much autonomy do agents have?

Accepted Answer

You decide. An approval broker sits between every agent and every action. High-risk actions always require your approval. In Copilot mode you approve everything; in Autopilot mode low-risk tools run on their own and you review the results; in Custom mode you choose which actions need sign-off, per agent and per tool. Confidence scores are recorded and shown to inform those thresholds, and they never auto-execute an action.

Question 4

How does the audit trail work?

Accepted Answer

Every agent action is logged with: what happened, why (the agent's reasoning), who triggered it, the confidence score, and the result. Entries are linked in a SHA-256 hash chain, so if any entry is tampered with, the chain breaks. You can verify integrity via the API or the audit log viewer.

Question 5

Can I undo what an agent did?

Accepted Answer

Yes. Every agent mutation captures a snapshot of the entity before and after the change. One-click rollback reverts to the pre-mutation state. The rollback itself is logged in the audit trail. This works for tasks, articles, lessons, courses, learning hubs, question banks, and questions.

Question 6

Can I run Neuphlo on my own infrastructure?

Accepted Answer

Yes, on every plan including free. The self-hosted bundle runs via Docker Compose with Postgres, Redis, PowerSync (local-first sync), LiveKit for voice/video, and Ollama with Llama 3.1 auto-registered as the default AI provider. Uploads go to your own S3-compatible bucket (AWS, Cloudflare R2, DigitalOcean Spaces, MinIO) or a local volume, and Neuphlo never proxies them. An interactive install wizard walks you through SMTP, OAuth, storage, and license choices; a one-line curl path takes sane defaults if you'd rather not answer prompts. Enterprise adds Helm charts, SSO, and SLAs.

Question 7

Can I use my own LLM instead of a cloud API?

Accepted Answer

Yes, on both cloud and self-hosted deployments. Connect your own model via a runner, or use Ollama, vLLM, or any OpenAI-compatible endpoint as a provider. The system also supports Anthropic, OpenAI, Google Gemini, and OpenRouter out of the box. You can configure different providers per workspace and route different agents to different models.

Question 8

Is Neuphlo designed for regulated industries?

Accepted Answer

The product is built with regulated sectors in mind. The tamper-evident audit trail is designed to support EU AI Act Article 12 logging requirements. EU-only data residency and crypto-shredding architecture support GDPR right-to-erasure workflows. Self-hosted deployment removes third-party data processing concerns entirely. Formal compliance certifications are on our roadmap. Today we provide the technical infrastructure and artifacts that compliance processes require.

Question 9

What happens when I reach the Starter plan limits?

Accepted Answer

You'll receive a notification when you're approaching your limits. You can upgrade to Solo (single seat) or Plus (for teams) at any time to unlock unlimited capacity. Your existing data, agents, and workspace preferences remain unchanged.

Question 10

How do Inbox seats work?

Accepted Answer

The Inbox is Neuphlo's shared customer support inbox, where email and messenger conversations are worked by your team and AI agents together. Inbox seats are an add-on to the Plus plan at €9.99 per seat per month (€99.99 per seat per year on annual billing), billed on the same interval as your plan and prorated. Buy the number of seats you need and assign them to team members under Preferences → Users. AI agents never need a seat. They're included free and don't count toward your seat total.

Question 11

Can I connect my own MCP servers?

Accepted Answer

Yes, on every plan. Add any remote MCP server over HTTP or SSE, or paste a standard mcpServers JSON config to import several at once. Neuphlo discovers the server's tools on connect and namespaces them per server, so nothing collides. You then choose which of those tools each agent is allowed to call, and every external MCP tool call stops for approval before it runs. Auth tokens and custom headers are stored with the server record, never returned to the browser and never placed in a model prompt. On cloud, servers must be reachable over public HTTPS; self-hosted instances can reach MCP servers on your own private network. Local stdio servers that launch a command on the host are not supported.

Question 12

Can agents work on files and run code on my own machine?

Accepted Answer

Yes, via the Neuphlo runner. Run npx @neuphlo/runner in the project you want worked on, with the CLI you already use (Claude Code, Codex, Gemini CLI, or Cursor), expose it over a tunnel, and add the URL under Preferences → AI → Providers. Agents then work in that directory: reading and changing files, running commands, running tests, using the developer tooling already installed there. Your code never leaves the machine, and you do not hand Neuphlo a model credential: the runner uses whatever your CLI is already authenticated with, subscription login or API key, whichever you set up when you installed it. Workspace actions such as creating a task or updating a page still execute on Neuphlo under the workspace identity of whoever started the run, with the permissions they have in the app, so a runner cannot widen its own access. Each CLI is launched into a throwaway home directory containing only Neuphlo's callback bridge, and callbacks are signed with a token that expires in fifteen minutes.

Question 13

How do source control integrations work?

Accepted Answer

Connect your GitHub or GitLab account via OAuth, including self-hosted GitLab instances. Once connected, agents can create branches from tasks, open PRs, monitor CI/CD status, and link commits to work items. All repo actions are logged in the audit trail.

Question 14

Can I import data from Jira, Linear, or other tools?

Accepted Answer

We're building import tools for popular platforms. Contact help@neuphlo.com for assistance with data migrations.

Question 15

Where is my data hosted?

Accepted Answer

Cloud deployments: all data is hosted in European data centres (Germany and Finland) on Hetzner infrastructure. No data leaves the EU. Self-hosted deployments: data stays on your infrastructure. Both options available on every plan.

AI agents run your projects. You make the decisions.

Every action audited. Every decision yours.

Nova drafts the plan

Approve, edit, or roll back

Signed to the audit trail

Your team. Your agents.

Brief an agent. Watch it work.

Work your agents actually manage.

Docs that maintain themselves.

Getting started with Task Management

Channels tied to the work.

Support on the leash you choose.

QA your agents run.

You set how much autonomy each agent gets.

Copilot

Autopilot

Custom

Trust through transparency, not promises.

Tamper-evident audit trail

One-click rollback

Confidence scoring

Run It Your Way

Cloud (EU)

Self-Hosted

Enterprise

Built for the teams that run the work.

Engineering Teams

Product Teams

Teams That Care About Accountability

Built with Compliance in Mind

EU Data Residency

Tamper-Evident Audit Trail

Human Override on Everything

Self-Host Option

Every plan includes AI agents. Enterprise adds governance.

Starter

Solo

Plus

Enterprise

Now on Mac, iPhone & iPad

Questions, answered.