Notion AI combines workspace search, writing, meetings, documents, and agents. The new developer surfaces make its automation infrastructure easier to discover and troubleshoot without leaving the workspace.
OUR VERDICT
A useful operational improvement for teams building Notion automations
Central access to deployments, connections, tokens, and run logs should make development easier. It also concentrates sensitive controls, so workspace roles, secrets, logs, and least-privilege access need deliberate configuration.
What changed
A new Developer section in Notion settings provides an overview of deployed Workers, connections, personal access tokens, and logs for Worker runs. Developer features can be enabled from workspace settings.
The developer bar
Notion also introduced a developer bar that can expose identifiers for pages, databases, blocks, workspaces, and users. Developers can copy an individual ID or the full API object. This can shorten debugging and integration setup, while increasing the need to avoid sharing objects that reveal workspace structure or sensitive metadata.
Why it matters for agents
Workers can run custom code for synchronization, custom-agent tools, and webhook automation. Faster access to run logs should help teams distinguish model errors from authentication, permission, API, or code failures. Logs should be treated as potentially sensitive because prompts, object names, payloads, or errors may contain business context.
Security checklist
Limit who can enable developer features, create tokens, deploy Workers, inspect logs, and add connections. Store secrets outside source code, rotate credentials, scope integrations narrowly, review outbound domains, document deletion and retention, and require approval for external or destructive actions.
How to evaluate it
Build a read-only Worker in a test workspace. Confirm permission boundaries, logging, failed-run visibility, token revocation, connection removal, object copying, and administrator oversight. Then test a write action behind explicit human approval.
Bottom line
The developer sidebar and bar make Notion’s agent and API ecosystem more manageable. They do not reduce the underlying risk of tokens, connected services, or executable Workers, so governance should expand with convenience.
See the Notion AI review, agent evaluation guide, and AI privacy review guide.
August 2026 update: clearer model and developer controls
Notion added a Developer section to the workspace sidebar for Workers, connections, personal access tokens, deployments, and run logs. A new developer bar can copy page, database, block, workspace, and user IDs or the full API object.
Notion also simplified its AI model selector with a shortlist for demanding tasks, scorecards comparing speed, intelligence, and cost, pinned favorites, and an effort control for more thorough responses.
Governance checks
- Restrict access tokens and developer features by role.
- Review Worker logs, connected-account scopes, credit usage, and failure handling.
- Document which model is approved for each data class and workflow.
- Confirm Business or Enterprise requirements for relevant controls.
Compare the best AI automation tools and review AI tool privacy before connecting data.
Our editorial read
Most valuable when notes, projects and team knowledge already live in Notion.
Best fit
Teams that want drafting, summarisation and lightweight agents close to their existing workspace context.
Where we would hesitate
Do not assume workspace context is automatically clean or permission-safe; review pages and sharing boundaries.
Decision rule
Choose Notion AI for workspace proximity; choose a general assistant when cross-app research or advanced coding is the priority.
Evidence note: This judgment synthesises the linked official documentation, release notes and product context. It is not a claim that AI Tools Manual independently benchmarked every tool.
Official source check
Official sources used for product facts: www.notion.so, www.notion.so. These links support named features, access, plan or policy statements; they are not independent benchmark results.
The links above are included because this page names or compares these products. They support the documented product facts only; the editorial judgment, workflow advice and limitations remain our own and are not presented as independent test results.
Notion connects meeting notes, agents, and credit tracking
Notion’s release notes describe Custom Agents that can be shared from the Share menu, AI Meeting Notes that can trigger a Custom Agent after a summary is complete, and Workers usage appearing in the credits dashboard. These changes connect meeting capture, workspace permissions, and automation economics.
Start with a reversible trigger: create a meeting summary, ask the agent to draft a follow-up, and require a human to approve any ticket, email, or database change. Inspect the credit record and audit trail. Do not assume a shared agent has the same access as its creator, and do not add confidential meeting material until permissions and retention are understood. See our knowledge-work shortlist for alternatives.
Primary source: Notion release notes. Verify credits, plan access, sharing permissions, and connected-app scope.
Notion 3.7: reusable team skills move from prompts to shared workflow
Notion’s September 15, 2026 release describes a shift from one-off prompting toward reusable team instructions. A workspace can keep shared AI skills in a library, organize them in a Notion database, and let teammates run a skill from an Agent chat. The release also says a skill can be exported as a SKILL.md with approved supporting files for tools such as Claude Code, Codex, Cursor, Gemini, and Grok. That makes Notion more useful as a place to maintain team operating guidance, not only as a place to ask an assistant questions.
The same release adds Custom MCP connections, including examples such as GitHub and Amplitude. Notion says these connections are in beta for Business and Enterprise and require confirmation before a connected tool is changed. This is a meaningful distinction: the feature can bring work context into an agent, but it is not a blanket permission to act on every connected system.
Who benefits, and where to be cautious
Best fit: teams that repeat the same review, reporting, onboarding, or document workflows and want instructions maintained in one shared location. It is less compelling for a solo user who only needs occasional free-form prompting, or for teams that need a fully independent knowledge base outside Notion. Cross-agent export is useful, but the organization still needs to manage versions, permissions, and the receiving tool’s own behavior.
Our assessment is that the most practical improvement is the reusable-skill library and portable SKILL.md format. Start with one low-risk workflow, write down its expected inputs and output, share it with a small group, then check whether the result is consistent before expanding it. For MCP connections, begin read-only where possible and keep approval prompts enabled; a connector’s presence should not be treated as proof that its results are complete or that an action is safe.
Notion’s release page also reports faster AI Search, but we have not independently benchmarked that claim, so it should be treated as a vendor statement rather than a measured result by AI Tools Manual. For a broader assistant comparison, see our ChatGPT review and Claude review. Evidence note: this update is based on Notion’s official 3.7 release notes; feature availability and plan eligibility should be checked in the current workspace. Next review: October 22, 2026.