PromptQL Tag vs Claude Tag
Your company's own AI. Any model. One shared brain.
Claude Tag brings Claude into Slack. PromptQL Tag gives your whole company its own AI across Slack and Microsoft Teams - with your brand, your models, your shared brain, and your users' permissions.
PromptQL Tag
Multiplayer AI coworker you can use in Slack and Microsoft Teams. Acts with permissions of the real human tagging it to read context and access data securely. It builds shared context in a transparent, portable wiki.
Runs any model and can be run in your own cloud.
Available in
Slack
TeamsClaude Tag
Anthropic's @Claude inside Slack. Access to tools and data requires admin setup per Slack channel. Memory is scoped to each Slack channel separately. Managed entirely by Anthropic.
Code, messages & data hosted and managed by Anthropic.
Available in
SlackChoose PromptQL Tag if you want less admin overhead to connect and share data sources, a shared knowledge and skills repository that stays portable across any LLM, and to host your data and your team's confidential conversations in your own cloud, integrated with your own security model.
Choose Claude Tag if you want to go all-in on Anthropic: you only foresee using Anthropic's models, and you want Anthropic to govern your shared memory, host your team's confidential conversations and data artifacts, and be your access-control registry for entitlements.
- ✓
Ownership: PromptQL Tag is your AI - your name, your models, your context, your security rules; Enterprise runs in your own cloud. Claude Tag is Claude, hosted by Anthropic.
- ✓
Memory: PromptQL Tag builds one shared, org-wide brain that compounds across Slack and Microsoft Teams. Claude Tag keeps memory siloed per Slack channel.
- ✓
Permissions: PromptQL Tag acts as the real user who tagged it, with their per-user permissions. Claude Tag acts as a shared per-Slack-channel bot identity.
- ✓
Models: PromptQL Tag runs any model plus image/video/audio generation. Claude Tag runs Anthropic models only.
- ✓
Both are SOC 2 and work in Slack. PromptQL Tag also works in Microsoft Teams; Claude Tag is Slack-only.
Feature comparison
Last updated: July 4, 2026
| Dimension | PromptQL Tag | Claude Tag (Anthropic) |
|---|---|---|
| Multiplayer features | ||
Q: Which is better for team collaboration in Slack, and does either one work in Microsoft Teams - PromptQL Tag or Claude Tag? Both work in Slack. PromptQL Tag also works in Microsoft Teams, while Claude Tag is Slack-only. In Slack, both work in public channels, private channels, and DMs, but PromptQL Tag also supports external/customer channels and a genuinely shared cowork session where multiple people drive the same AI thread together. | ||
| Shared cowork session | ✓Yes - one shared AI thread everyone drives together. | ✗No - only a read-only view of Claude's session. |
| Works in Microsoft Teams | ✓Yes | ✗No - Slack only |
| Works in Slack public channels | YesEvery public channel | YesEvery public channel |
| Works in Slack private channels | YesEvery private channel | YesIsolated per-channel identity |
| Works in Slack DMs | YesIn DMs | YesOn the user's own account |
| Works in Slack external / customer channels | ✓Yes - guests scoped to a slice of context | ✗No |
| Context - the shared brain | ||
Q: Does Claude Tag share memory across Slack channels? Can PromptQL Tag share one brain across Slack and Microsoft Teams? Claude Tag does not - its memory is siloed per Slack channel, so the same facts are re-learned in every channel. PromptQL Tag builds one shared, org-wide brain across Slack and Microsoft Teams: a Wikipedia-style wiki anyone can read, navigate, and edit, with revision history and review, where rooms share overlapping slices of the same knowledge. | ||
| One shared, org-wide brain | ✓Yes - one brain shared across Slack and Microsoft Teams | ✗No - memory siloed per Slack channel; same facts re-learned in each |
| Knowledge compounds across teams | ✓Yes - teams in Slack and Microsoft Teams add to the same brain | ✗No - knowledge fragments along Slack channel walls |
| Context model | ✓Wikipedia-style - human-readable & editable | ✗Black-box per-Slack-channel notes |
| Revision history & review | ✓Yes - change history + notifications to review/revert | ✗No version history; no pre-save review |
| Overlapping scopes across rooms | ✓Yes - Slack and Microsoft Teams spaces share the right slice | ✗No - Slack channels can't share a controlled slice |
| Integrations | ||
Q: Who sets up tools and data access - do you need admins and developers? In PromptQL Tag, any user can self-serve connect data and APIs they own, and the AI coworker sets the connector up conversationally. In Claude Tag, integrations and tool access are set up per Slack channel by admins and developers together via manually configured "Access bundles." | ||
| Who sets up connectors and tool access | ✓Any user, for data/APIs they own | ✗Admins and developers together |
| Conversational connector setup | ✓Yes - coworker maps the API & wires up auth | ✗No - manual per-Slack-channel "Access bundles" |
| Security | ||
Q: Whose permissions does the AI coworker use - the user's or a bot's? PromptQL Tag acts as the real person who tagged it: scoped API tokens for custom APIs, per-user OAuth for OAuth2 apps, and per-user row-level security for databases - so audit logs show the actual human. Claude Tag acts as a shared per-Slack-channel service-account identity (per-user OAuth only in Slack DMs), and external systems log the bot. | ||
| Whose permissions the coworker uses | ✓The user who tagged it | ✗A shared per-Slack-channel bot identity |
| Per-channel blast-radius isolation | YesVia room scopes | YesEach Slack channel isolated by design |
| Custom APIs / agents | ✓Scoped API tokens | ✗Shared Slack channel credential |
| OAuth2 APIs | ✓Per-user OAuth | ✗Shared account (per-user only in Slack DMs) |
| Databases | ✓Per-user row-level security (RLS) | ✗Shared service account; no per-user RLS |
| Audit attribution | ✓The real human | ✗The bot (human only in Anthropic's log) |
| AI models | ||
Q: Which AI models can each one use? PromptQL Tag runs any model - Claude, GPT, Gemini, DeepSeek, open-weight - chosen per channel or per task, and can generate images, video, and audio in-thread. Claude Tag runs Anthropic models only, with no non-Anthropic model choice and no image/video/audio generation. | ||
| Model choice | ✓Any model (Claude, GPT, Gemini, open-weight) | ✗Anthropic models only |
| Image / video / audio generation | ✓Yes | ✗No |
| Branding & deployment | ||
Q: Can the coworker carry your company's brand and run in your own cloud? PromptQL Tag posts under your company's own name and avatar, and Enterprise runs in your own cloud (VPC/BYOC) with SSO. Claude Tag posts as "Claude" and is hosted only by Anthropic (not available to Zero-Data-Retention orgs). | ||
| Your brand & name | ✓Yes - your name & avatar | ✗No - posts as "Claude" |
| Runs in your own cloud | ✓Yes (VPC / BYOC) + SSO | ✗No - Anthropic-hosted only; no ZDR |
| Pricing shape | –Usage-based | –Usage-based, no per-seat |