> For the complete documentation index, see [llms.txt](https://docs.sprinto.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.sprinto.com/product-updates/july-2026.md).
# July 2026
Headlined by Sprinto MCP opening your compliance data to AI agents and Playbooks turning multi-step processes into repeatable, auditable AI-run workflows, plus evidence excerpts in AI-generated answers and two Agent Mode upgrades (viewable artifacts, and notifications with follow-up suggestions). Here’s what’s new.
## Evidence Excerpts in Security Questionnaire Auto-Answers
AI-generated questionnaire answers now show the exact excerpt from the source evidence they were built from — not just a link to the full document.
### What’s new
* Inline excerpt/quote shown alongside each AI-generated answer, clearly marked as AI-sourced
* Supports multiple excerpts per answer when more than one source chunk was used
* Available for AI-generated answers only (manual answers are unaffected)
### Why it matters
* Lets reviewers validate an auto-answer without opening the full source document — a meaningful time save on 30+ question questionnaires
* Builds trust in AI answers by showing the “proof,” not just the conclusion
**Availability:** This is an on-demand feature. Raise a request to get this feature enabled for your organisation.
## Sprinto MCP
Sprinto now has a Model Context Protocol (MCP) server, letting MCP-compatible AI clients — ChatGPT, Claude, and other agent platforms — connect directly to your Sprinto organisation and query or act on your compliance data on your behalf.
### What’s new
* **Connect via OAuth.** Authenticate through Sprinto’s own login; every action the connected AI takes is attributed to and scoped to that authenticated user, admin-gated.
* **Read tools.** You can query across controls, checks and monitors, frameworks, policies, evidence, risks, vendors, audits, incidents, access reviews, staff and devices, cloud infrastructure, pentests, training, and Trust Center. Queries return records, full details, status and readiness, and reports. Readiness counts take approved exceptions into account, so they match what Sprinto shows. You can also search your Knowledge Hub with citations and download underlying documents through time-limited links.
* **Write tools.** You can write across controls, checks and monitors, frameworks, policies, evidence, risks, vendors, audits, incidents, access reviews, staff and devices, pentests, security questionnaires, and Trust Center.
* **Org-level kill switch.** Admins can disable MCP access for their entire organisation from Sprinto settings.
* **Claude Connectors Marketplace listing.** A curated, explicit tool set (without the generic catch-all tools) is available for Claude’s Connectors Marketplace.
### Why it matters
* Lets your team (or your own AI tooling) pull live compliance answers and status into whatever AI assistant you already use, instead of logging into Sprinto separately.
* Batch questionnaire answering turns a manual, multi-hour RFP/security-review response into a single request.
* The org-level kill switch means IT/security teams retain a hard stop over AI-agent access at any time.
**Availability:** Live in production — multiple customers are already connected via ChatGPT and Claude. An org-level kill switch exists; per-connector-level disable is not yet available.
## Playbooks
Playbooks are Sprinto’s reusable, multi-step AI-agent workflows for driving a specific compliance or security outcome end-to-end — like onboarding and assessing a new vendor, or drafting a policy — instead of doing every step by hand.
### What’s new
* **Playbook templates.** Each Playbook has a charter (goal, scope, guardrails) and an ordered set of steps, each assignable to a specific AI agent with its own retry/condition logic.
* **Human-in-the-loop review.** Runs can pause for interactive approval forms right inside the run’s activity feed, with accept/skip actions on each gate.
* **Run visibility.** A live timeline of steps, an activity log, generated artifacts, and a run summary for every execution; runs can be paused and shared with teammates.
### Why it matters
* Turns a recurring, multi-step compliance process into a repeatable, auditable, AI-run workflow instead of an ad hoc checklist.
* Human-in-the-loop gates keep judgment calls (e.g. approving vendor communication) with a person, while the AI handles the repetitive steps around it.
**Availability:** Live, under “AI Toolkit → Playbooks.”
## Agent Mode: Artifacts
When Sprinto’s AI agent produces a document-style result — a policy, a report, charts, tables, graphs or another generated artifact — it now renders as a live, viewable document in chat instead of a plain link or wall of text.
### What’s new
* **In-chat document viewer.** Generated artifacts render as a PDF-style viewer, either as a compact preview card inline in the conversation or in a resizable side panel next to the chat.
* **Live-rendered drafts.** If the agent authored the document itself and it isn’t backed by a stored file yet, the viewer shows a live rendered page of that content rather than a placeholder.
* **Zoom, copy, print, download.** Standard document controls are available directly from the panel, including a one-click PDF download.
* **Expiry handling.** Download links are time-limited; if one expires, you get a clear message and can refresh or re-ask instead of hitting a broken link.
### Why it matters
* Lets you review and act on AI-generated documents (policies, reports) without leaving the conversation or waiting on a separate export step.
* In-place editing means small corrections don’t require re-generating the whole document from scratch.
## Agent Mode: Notifications & Follow-Up Suggestions
Two related additions to Agent Mode: a live status indicator for your AI conversations, and suggested next-step chips after the agent responds.
### What’s new
* **Live status pill.** The sidenav now shows a real-time indicator for your agent conversations — a “thinking” animation while an agent is actively working, and an unread/toast notification when a conversation is ready for your review or needs your approval.
* **Smart toast behaviour.** You won’t get re-notified about a conversation you already have open, and only genuinely new activity triggers a toast.
* **Follow-up suggestion chips.** After an agent responds, it may suggest one or more clickable next steps — either a follow-up question or a specific action tied to a record (e.g. “Assign an owner,” “Review \[Vendor Name]”) — shown on the latest message in the thread.
### Why it matters
* Keeps you from having to babysit a conversation waiting for the agent to finish — you’re notified the moment it needs you.
* Follow-up chips turn a single AI answer into a guided next step instead of a dead end, reducing the “now what” moment after a response.
## Quick Updates
Smaller improvements and fixes shipped this month:
* **Vendor Discovery:** “Last Accessed” date is now shown against discovered/suggested apps, making it easier to spot unused apps and shadow IT.
* **Vulnerabilities:** Added an export option for vulnerability and asset data, so it can be cross-checked against external trackers.
* **Task Dashboard:** Send a reminder for one specific failing task instead of nudging every assignee at once, with the option to CC the owner’s manager.
## Wrapping Up
That’s July. As always, if you’d like a deeper walkthrough of any of these, reach out to your CSM or drop a note in our support channel.
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.
Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:
```
GET https://docs.sprinto.com/product-updates/july-2026.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.