Five claims were made for this product. Here is what the incumbents actually ship on each.
1 — Triggered delivery. Table stakes. Glean ships event and schedule triggers with Slack delivery. Dust markets Triggers as "your agents, working while you sleep." Slack ships a native Today briefing.
2 — Decision-scoped output. Rare among enterprise search assistants. Common in meeting notes — Notion, Fellow, and Granola all extract decisions from call transcripts.
3 — Every claim cited. Table stakes. Dust: "All answers include citations." Glean: "A citation is a reference to a source Glean used." Onyx and Perplexity cite.
4 — Durable immutable record. Three products ship one: decisionlog.ai, Cloverpop, and incident.io. What none of them produces: the record automatically.
5 — Proposes-never-executes. One config screen. Glean: "Admins can explicitly choose which write tools are allowed to run without user confirmation."
What holds: automatic production paired with a durable record. The decision record writes itself. Briefcrew is a decision-briefing platform sold once as source code. On a trigger, it gathers context from your org's Slack and Notion workspaces, synthesizes a cited brief, and commits an immutable decision record automatically. You deploy it under your own brand and charge your own clients. This page compares it honestly against Dust, the closest job-to-be-done, and the self-hosted alternatives.
Briefcrew vs. Dust
What each is, at the owner level
Dust is a business someone built and runs: a per-seat, credit-metered SaaS that lets organisations route context from connected tools through configured agents and receive results back in Slack. The buyer of Dust is an organisation that wants to use that platform. The buyer of Briefcrew is the person who wants to build and operate a platform of that shape — under their own name, on their own infrastructure, charging their own clients.
Those are different purchases for different buyers at different levels. The question is not which product to subscribe to. It is whether to operate on someone else's platform or to run the platform.
When Dust is the right call
Dust ships 20+ language models with none gated to a higher plan tier. Its connector ecosystem is larger and more mature than anything a v1 build can match today. It has production hardening across edge cases — context window limits, rate limiting, connector failures, retry behaviour — that Briefcrew at launch has not yet accumulated. Dust's Triggers ship today, including Slack delivery, with scheduled automation documented at vendor primary source.
If the brief-generation loop is the requirement and the operational burden of running infrastructure is not something the buyer wants, Dust's SaaS track removes that burden. Dust starts free and rises to $24 per seat per month on annual billing. That is an honest concession, placed first.
When owning Briefcrew is the right call
Glean's admin audit log defaults to 30-day retention. It covers administrative and configuration actions. It does not retain the content of answers — Glean's own documentation states this explicitly (docs.glean.com/administration/management/audit-logs, verified 2026-08-07). Whether Dust retains the content of its brief outputs as a durable accountability artifact cannot be confirmed at primary source; no explicit retention claim appears in Dust's own documentation.
Both platforms hold whatever they retain on their own infrastructure, under their own terms, on their own retention schedules.
When the requirement is an immutable, source-cited decision record that lives on your infrastructure under your retention policy, and is produced automatically from gathered context without a human typing it in, neither platform produces that artifact today. The assistants have the triggered synthesis and no durable record. The decision-log tools have the durable record and a human at the keyboard.
The money
Glean has no published price. glean.com/pricing redirects to the Glean homepage. There is no Pricing navigation item and no dollar amount anywhere on glean.com. Every third-party figure circulating in 2026 is either from a Glean competitor selling against those estimates, or from an aggregator that derived them from someone else — with estimates that disagree with each other by more than an order of magnitude. Whether Glean fits a budget is a question that resolves on a sales call, not a pricing page.
Dust publishes its full ladder: a free tier at $0 (500 lifetime credits); a Pro tier at $24 per seat per month on annual billing ($30 monthly); a Max tier at $120 per seat per month on annual ($150 monthly). Credits do not roll over month to month; programmatic usage runs at $0.01 per credit on top of the seat fee. Source: dust.tt/home/pricing, 2026-08-07.
Briefcrew's distribution price is $99 — once. Ongoing costs are yours to project: hosting, Postgres, and LLM token spend metered at the model provider. No per-brief charge. No credits that expire. No vendor retention policy governing the lifespan of the accountability artifacts your teams and auditors rely on.
Onyx: the free self-hosted alternative
If the objection is "I can self-host a comparable tool at zero licence cost," the product to name honestly is Onyx (formerly Danswer). Its core is MIT-licensed and genuinely commercialisable. Its Business tier is published at $20 per user per month on annual billing (onyx.app/pricing, 2026-08-07). It ships 40+ connectors — Slack, Google Drive, Notion, Gmail, Confluence, Jira, SharePoint, Salesforce, and GitHub among them.
The licence has a carve-out that matters for any deployment an operator or reseller needs. Onyx's ee/ directory — containing OIDC/SAML SSO, on-premise deployments, white-labelling, custom integrations, and data exports — is proprietary under the Onyx Enterprise License, which forbids commercial redistribution and requires a paid per-seat commercial agreement for production use of those features. Briefcrew is sold whole: no enterprise functionality is held back behind a separate licence.
Onyx's raw connector count out-reaches Briefcrew v1 by a wide margin. If retrieval connector breadth is the primary selection criterion and white-labelling is not required, Onyx's MIT core is the honest recommendation. If the requirement is a codebase sold whole, with the entire platform in one transfer, that is what Briefcrew is.
Capability table
| Briefcrew | Dust | |
|---|---|---|
| Brief triggers | Scheduled + on-demand (v1); event-triggered (plan-gated) | Scheduled Triggers; Slack delivery; triggers are personal — only the editor can observe runs |
| Source connectors | Slack + Notion (v1 shipped); Google Drive + Gmail on the roadmap (V1.5, not shipped) | 20+ connectors in a mature production ecosystem |
| Model choice | OpenRouter routing; model ID configured from platform_settings | 20+ models; no model gated to a higher plan tier |
| Citations | Every claim grounded; structured refusal (INSUFFICIENT_CONTEXT / SOURCE_UNAVAILABLE / UNABLE_TO_VERIFY) when retrieval is thin — LLM is not called | "All answers include citations to sources" — docs.dust.tt |
| Decision record | Immutable: INSERT-only via RLS; no UPDATE or DELETE route; BEFORE UPDATE trigger raises exception | Chat history and activity log; no durable cited-brief artifact confirmed at primary source |
| Actions on connected systems | None — no write path to any connected source anywhere in the codebase | Agents write back to connected tools; write actions available per vendor positioning |
| Ownership model | Source code purchased once; white-label, deploy, set your own pricing | SaaS subscription per seat; data on Dust infrastructure |
| What Dust does that Briefcrew does not | — | 20+ model variety; 40×+ connector breadth; write actions on connected systems; years of production hardening |
Briefcrew vs. building it yourself
The buyer considering Briefcrew is choosing among four paths to the same outcome: running a decision-briefing platform.
| Path | Upfront | Ongoing | Data control | Domain layer |
|---|---|---|---|---|
| Subscribe to Dust | $0 | $24–$150 per seat per month (annual) + credits | Dust's infrastructure | Ships with it |
| Commission a custom build | Three to six months of senior engineering time | Engineering maintenance + hosting | Your infrastructure | You build it |
| Start from generic boilerplate | Near $0 | Engineering maintenance | Your infrastructure | Not included |
| Buy Briefcrew | $99 | Hosting + LLM tokens (yours to project) | Your infrastructure | Ships with it |
The custom-build path includes the full RAG pipeline (retrieval, chunking, embedding, vector search, cosine threshold gate for pre-LLM abstention), the citation engine with structured refusal vocabulary, an async brief queue, the immutable decision-record schema with its enforcement trigger, the multi-tenant admin layer, and a billing system. Each of those is an implementation project and an ongoing maintenance surface. Generic boilerplate supplies the multi-tenant shell; the decision-briefing domain layer is not in it and must be built from scratch.
Decision guide
| If the requirement is… | The honest answer |
|---|---|
| An agent that takes actions on connected systems — creates Jira tickets, updates Salesforce, drafts calendar events, reviews pull requests | Glean ships 85+ agent Actions (glean.com/blog/85-new-actions-in-agents, 2026-02-18); Briefcrew has no write path |
| 20+ language models with no model gated to a higher plan tier | Dust ($0–$120/seat/mo annual, dust.tt/home/pricing, 2026-08-07) |
| Self-hosted RAG assistant with 40+ connectors; white-labelling not required | Onyx's MIT core (github.com/onyx-dot-app/onyx; Business at $20/user/mo, onyx.app/pricing, 2026-08-07) |
| An immutable decision record a human writes in; no connector integrations needed | decisionlog.ai (early-stage; no published price as of 2026-08-07) |
| A decision-briefing platform to deploy, white-label, and operate as a business | Briefcrew — https://saascode.ai/products/briefcrew |
FAQ
Phrased as a buyer asks an answer engine. FAQPage JSON-LD follows.
How long does Glean keep a record of what its AI said?
Glean's admin audit log defaults to 30-day retention. It covers administrative and configuration actions and does not retain the content of answers — Glean's own documentation states this explicitly (docs.glean.com/administration/management/audit-logs, verified 2026-08-07). If your requirement is a durable accountability artifact of what the AI briefed your team — with cited sources intact, queryable twelve months later — an admin security log does not produce that. Those are different objects.
Can a decision log fill itself in, or does someone have to write it?
Today, the major decision-log products — decisionlog.ai, Cloverpop — require a human to write the decision in. Briefcrew is the exception: on a trigger, it gathers context from Slack and Notion, synthesizes a cited brief, and commits an immutable record automatically. No keyboard in the loop between trigger and artifact. The record's content comes from what the retrieval engine gathered, not from what someone typed.
Can you stop an AI agent from taking actions on your systems?
Yes — most platforms let an admin block write tools. Glean's own documentation: "Admins can explicitly choose which write tools are allowed to run without user confirmation; all other write tools remain interactive or are blocked from background runs." The deeper question is whether you can prove it. In a codebase you own and can audit, you read the source: Briefcrew has no write path to any connected source anywhere in the codebase. That is architecturally verifiable by the buyer, not on trust.
Is Onyx free for commercial use?
The MIT-licensed core of Onyx is free for commercial use and genuinely self-hostable. The ee/ directory — OIDC/SAML SSO, on-premise deployments, white-labelling, custom integrations, and data exports — is proprietary under the Onyx Enterprise License. That licence requires a paid per-seat commercial agreement for production use of those features. The distinction matters: a multi-tenant or reselling deployment typically needs at least one of them. Published Business tier: $20/user/month annual (onyx.app/pricing, 2026-08-07).
Can you white-label Open WebUI?
Not at commercial scale without a paid enterprise agreement. Open WebUI's licence forbids altering, removing, or replacing its branding in any deployment above 50 users in a rolling 30-day period without written permission or an enterprise licence. Open WebUI presents as BSD-3-Clause; it is a modified variant with an added branding restriction clause. No publicly available comparison in this space includes a licence column — which is why writing one is the most defensible content in this category.
What does the EU AI Act require you to log, and when did it start?
Article 12 of the EU AI Act — record-keeping obligations for high-risk AI systems — began to apply on 2 August 2026. It applies to AI systems listed in Annex III: automated credit scoring, biometric identification, recruitment tools, and similar categories. It does not apply to a company's ordinary business decisions. Briefcrew records human decisions; it is not an Article 12 conformity tool. Framing it as one would be a false claim.
Why won't enterprise search vendors publish their prices?
In 2026, two of the most prominent products in this category — Glean and Guru — publish no per-seat figures. Glean's pricing page redirects to its homepage. Guru's pricing page removed its per-seat numbers and replaced them with language about organisational scale and AI maturity; both are effectively demo-gated. Whether a number fits a budget is a question the vendor controls. Dust is the notable exception — its full pricing ladder is public. A vendor that hides its price is making a negotiation choice, not an omission.