# Bloomfire — Keeps a persistent semantic memory of connected systems, so AI answers from governed company knowledge. > Source: AgentsIndex — https://agentsindex.ai/bloomfire (structured, researched, re-verified) Bloomfire is an enterprise knowledge platform that indexes connected systems such as SharePoint, Confluence and Salesforce into a continuously updated semantic memory. Its engine reasons over that memory and enforces content policies before an answer reaches an employee or an AI agent. **Best for:** Regulated enterprises grounding their own AI agents in governed internal knowledge **Look elsewhere if:** - you want to start without talking to sales - you need to cancel any time on every plan | Fact | Value | | --- | --- | | Website | https://bloomfire.com | | Category | AI enterprise search | | API | Yes | | Open source | No | | Price | Price on request | | Price source | https://bloomfire.com/pricing/ | | Price checked | 28 October 2026 | | MCP server | Yes | | Self-host | Partly | | Facts | 20 facts from 10 sources | | Facts checked | 28 October 2026 | Screenshot of Bloomfire: https://agentsindex.ai/img/1920/screenshots/bloomfire.webp, captured 29 October 2026 **What we checked** - Pages last read: 28 October 2026 - Facts checked: 20 facts (From 10 sources) ## What Bloomfire actually does Bloomfire began in 2011 as knowledge management software and now sells the layer underneath it: the Enterprise Intelligence Engine, which reads connected systems and keeps what it learns in what the vendor calls a Persistent Semantic Memory. The claim on the [AI Connect page](https://bloomfire.com/eie/ai-connect/) is that most AI-ready APIs are only a search endpoint, so the model receives a stack of documents and still has to work out what is current and who may see it. The product sits in two halves. The knowledge management app is where people author, moderate and search content, with Synapse answering questions conversationally and inside [Slack](https://bloomfire.com/integrations/slack/) and Microsoft Teams. The engine is headless, and its surfaces are sold as reasoning functions an agent calls: Synapse Active Intelligence for grounded answers, AI Signal Studio for triggers, Learning Studio for decision paths, Organization Simulator for pressure-testing a knowledge base. Content arrives through named connectors for SharePoint, Google Drive, Confluence, Salesforce, GitHub, Zendesk, Box, Zoom, Gong and Genesys, and goes back out to Amazon Q, Microsoft Copilot and Salesforce Einstein. A second group of connectors covers paid research subscriptions such as Mintel, Kantar, Euromonitor, Forrester and S&P Global, which points at the market research and insights teams Bloomfire has sold to for years. ## Agents call the engine over MCP, inheriting the caller's permissions Every engine surface is reachable over both a REST API and [Bloomfire's own MCP server](https://bloomfire.com/eie/ai-connect/), which the vendor describes as exposing those surfaces as callable tools. The page names Claude, Microsoft Copilot and Glean over MCP, ChatGPT through published function schemas, Gemini through function declarations, and any HTTP client through REST. Two details matter more than the list. Authentication is per user, so an agent acts with one person's rights and source system access controls still apply: if a user cannot open a file in SharePoint, an agent acting for them does not reach it. And access checks run during reasoning rather than as a filter on the output, so restricted concepts are excluded before an answer is assembled. The [REST API documentation](https://help.bloomfire.com/api/documentation/) is public and, per the vendor's [integrations FAQ](https://bloomfire.com/integrations/), covers searching, returning, creating and modifying posts from any tool a customer already uses. Intelligence Modules each ship with their own endpoints, so a developer can call a module directly or chain several into a multi-step agent workflow. ## Guardrails are policies with severity tiers, not reports Content Guardrails is the part of the platform with the most specific documentation. Policy families ship for credentials, PII and PHI, GLBA, confidentiality markers, deprecated terminology, language and link hygiene, and each policy carries its own severity and its own action: flag it, suppress it from search and AI, or quarantine it and notify an administrator. Four severity tiers map to effects a buyer can predict. Eligible content participates fully, Warning is advisory, Violation is flagged for human review while retrieval continues, and Critical or Blocked content is excluded from search and synthesis but stays visible in the health report. For credentials, PII and GLBA the vendor says detection is fully deterministic with no model calls, using pattern matching and checksum validation. Content arriving from a connected system can also be held in sandbox for admin review before it becomes searchable. Custom policies are added as a single JSON file, which is the line where configuration stops being point-and-click. ## Pricing is a quote, and the terms contradict the pricing page There is no rate card. The [pricing page](https://bloomfire.com/pricing/) is a contact form whose promise is a fixed price scoped by sales, stated as no consumption charges or user counts. No free plan and no trial appear anywhere on the site. The [terms of use](https://bloomfire.com/terms-of-use/) describe a different billing shape. A subscription specifies limits on items including the number of members, and exceeding them is charged at a per unit fee from the subscription or from Bloomfire's then current list price. Accounts are designated user accounts that cannot be shared, and sharing can trigger suspension or an invoice for the extra members. The worked example in the terms charges a subscriber for 75 additional members after exceeding the limit by 54. Neither the per unit fee nor the list price is published, so the overage is unpriced for a buyer reading the site. Subscriptions are paid in advance, are non-refundable, renew automatically, and require written notice at least 30 days before the end date to stop. ## SOC 2 Type II is the only audit, and ISO 27001 was dropped in 2022 The [trust center](https://bloomfire.com/trust-center/) is a Vanta-hosted page listing a SOC 2 Type II report dated February 2026, the NIST AI Risk Management Framework, policy documents available on request, and controls grouped by infrastructure, organizational, product and privacy. Subprocessors are named with their regions: AWS carries all hosting, with CloudConvert for document processing, Mailgun and SendGrid for email, and MongoDB Atlas holding community content and vector embeddings on private-only connectivity. ISO 27001 is a deliberate absence rather than a gap. The [data security FAQ](https://bloomfire.com/kms/data-security/) says Bloomfire ran both programs for a while, found substantial overlap, and in 2022 adopted SOC 2 as its single compliance program. Healthcare and EU claims are phrased as HIPAA-Ready and GDPR-ready, which is not the same as naming a signed business associate agreement or data processing addendum, and neither document is published. Identity handling is concrete: SAML 2.0 and OIDC for sign-in, SCIM for Azure Active Directory and Okta, and an [implementation estimate](https://bloomfire.com/scale-your-ai/) of under four hours for IT to connect both. ## Where it fits, and where it does not The buyer this is written for is a large regulated enterprise that already has content scattered across SharePoint, Confluence and a helpdesk, and that wants one governed layer its own agents can query. The company page claims 15% of the Fortune 100 as customers, and the guardrail families for GLBA and PHI point at banks, credit unions, insurers and healthcare administration. It is a poor fit for a small team. There is no way to start without sales, no published price, and the implementation is described as a phased program with onboarding, content migration and change management staff attached. Teams wanting to evaluate the engine first have a product tour and a demo request, not a sandbox. On-premise and air-gapped deployment is listed as an engine option, while the security FAQ says most deployments are multi-tenant cloud and stricter architectures are scoped case by case. A buyer whose requirement is on-premise should treat it as a negotiation rather than a product feature. Growth equity came from [Primus Capital](https://bloomfire.com/newsroom/investment-primus-capital/) in 2021, and the company is now led by CEO Philip Brittan. ## Bloomfire price and terms **Terms** - Sign up without sales: No ## For AI agents - MCP server: Yes - API: Yes Pricing page as we captured it: https://agentsindex.ai/img/1920/screenshots/bloomfire/pricing.webp, captured 29 October 2026 Read on https://bloomfire.com/pricing/, 28 October 2026. ## What we checked on Bloomfire ### What it costs to start **What the price counts** · Yes — Priced only on a quote: sales scopes the deployment and returns one fixed figure that the page says is not metered by consumption or by user count. > Get a fixed price based on your intended scope - no consumption charges or user counts > — https://bloomfire.com/pricing/ · Ready For A Price Quote? panel, first bullet, checked 28 October 2026 **Self-serve signup** · No — There is no signup path: the pricing page's only instruction is to contact the sales team, which then scopes and quotes the deployment. > Contact a member of our sales team and we will be able to discuss scope and provide you with a custom price for you. > — https://bloomfire.com/pricing/ · Ready For A Price Quote? panel, opening paragraph, checked 28 October 2026 **Contract** · Yes — Subscriptions run for a fixed term that renews by default, and each subscription carries named limits on members and items that cost extra once exceeded. > Bloomfire subscriptions automatically renew unless a proper notice of cancellation is completed at least 30 days prior to the end date of the current subscription. > — https://bloomfire.com/terms-of-use/ · Section 8, Cancellation and Termination, clause a.i, checked 28 October 2026 **Cancel any time** · Partly — Cancellation is possible but never free of a term: it takes written notice at least 30 days before renewal, and fees already paid are not refunded. > Subscriber can cancel a Membership Community, with at least 30 days notice, by e-mailing cancel@bloomfire.com specifying the Membership Community that you want canceled > — https://bloomfire.com/terms-of-use/ · Section 8, Cancellation and Termination, clause a.i, checked 28 October 2026 ### What it works with **Works with** · Yes — Named connectors cover SharePoint, Google Drive, Confluence, Salesforce, GitHub, Zendesk, Box, Zoom, Gong, Slack and Teams, plus outbound push to Amazon Q and Microsoft Copilot; the site ties none of them to a plan. > Bloomfire connects to the platforms your team already uses, from SharePoint and Google Drive to Confluence, Gong, GitHub, Salesforce, and more. > — https://bloomfire.com/integrations/ · Knowledge Connectors section, introductory paragraph, checked 28 October 2026 **Public API** · Yes — A public REST API is documented at help.bloomfire.com and covers searching, retrieving, creating and modifying posts; the engine's reasoning surfaces are reachable the same way. > Yes! You can access our api documentation here . Use this documentation to create links to search, return, create or modify posts from any tool you use today. > — https://bloomfire.com/integrations/ · Frequently asked questions, 'Do you have open API' answer, checked 28 October 2026 **MCP server** · Yes — Bloomfire runs its own MCP server exposing each engine surface as a callable tool, with per-user authentication so an agent inherits only that person's permissions. > Bloomfire's MCP server exposes the engine's surfaces as callable tools, so an agent reaches grounded knowledge through a protocol it already speaks rather than a custom integration. > — https://bloomfire.com/eie/ai-connect/ · Frequently Asked Questions, 'What is Model Context Protocol, and why does Bloomfire use it?', checked 28 October 2026 ### What it does on its own **Acts on its own** · Yes — AI Signal Studio agents monitor connected systems continuously and act once a threshold set by an administrator is met, without a person approving each firing. > Signals are autonomous AI agents that operate within clear rules, that you can set. They continuously monitor the connected systems and the knowledge within their scope. > — https://bloomfire.com/eie/ai-signals/ · AI Signal Studio introduction, opening paragraph, checked 28 October 2026 **Limits on what it may do** · Yes — Content Guardrails give every policy a severity tier and an action from advisory flag to hard block, and connected agents must check content against those policies before acting. > Policy flags work as AI guardrails, not just reports. Any agent or app connected through AI Connect can check content against your policies via API or MCP before acting on it. > — https://bloomfire.com/eie/content-guardrails/ · Access Control for Agents section, checked 28 October 2026 **No code needed** · Partly — Administrators configure preset guardrail policies and pre-built signals in the app, but a custom policy is authored as a JSON file and building on the engine is API or MCP work. > Easily add your own with a single JSON file. > — https://bloomfire.com/eie/content-guardrails/ · Configurable Policy Library section, checked 28 October 2026 ### What happens to your data **No training on your data** · Yes — Customer content and the semantic memory built from it are kept out of generative model training, and the memory is held separate from the language model rather than fine-tuned into one. > Your data and your PSM are never used to train or fine-tune a generative model. > — https://bloomfire.com/eie/ai-connect/ · No training on your content section, checked 28 October 2026 **Retention you control** · Partly — Administrators can schedule archiving and unpublishing of aging content and recover deletions for 30 days, but no customer-set retention window or purge control is published. > You can schedule publishing for a future date, as well as automatically scheduling archiving and unpublishing documents when they become out of date. > — https://bloomfire.com/kms/moderation-tools/ · Frequently asked questions on scheduling and archiving, checked 28 October 2026 **Data stays in your region** · Partly — Where content sits depends on the deployment: it can stay in the customer's own cloud storage or move into a Bloomfire tenant, and no region menu is offered to the buyer. > Depending on the configuration, data may remain on your existing cloud base storage or be uploaded to a Bloomfire platform tenant. > — https://bloomfire.com/kms/data-security/ · Frequently Asked Questions, 'Where is my data stored?', checked 28 October 2026 **Self-hosted** · Partly — Air-gapped and on-premise deployment is listed among the engine's options, while the security FAQ treats anything other than multi-tenant cloud as a case-by-case scoping conversation. > Multi-tenant, single-tenant, air-gapped, or on-premise. The engine needs access to your storage solution to build and maintain the memory, but your content stays where it already lives. > — https://bloomfire.com/eie/ai-connect/ · Deployment options section, checked 28 October 2026 **Open source** · No — The platform is proprietary: the terms reserve all software and intellectual property to Bloomfire and grant only a revocable license, with FFmpeg the single open component listed. > the Bloomfire Service and its materials, including software, images, text, graphics, illustrations, logos, patents, trademarks, service marks, copyrights, photographs, audio, videos and music, and all > — https://bloomfire.com/terms-of-use/ · Section 6, Our Proprietary Rights, checked 28 October 2026 **Take your data out** · Yes — Customers can take a backup of all their content before canceling, and report data exports to CSV for use in tools such as Tableau or Power BI. > You may generate a backup of all of your User Content prior to canceling your Membership Community. > — https://bloomfire.com/terms-of-use/ · Section 8, Cancellation and Termination, clause a.i, checked 28 October 2026 ### Whether a company can buy it **Security audits** · Yes — SOC 2 Type II is the named audit, renewed annually and shared under NDA; the trust center also lists the NIST AI Risk Management Framework, and ISO 27001 was deliberately dropped in 2022. > in 2022, Bloomfire has decided to adopt SOC 2 as our single security compliance program going forward. > — https://bloomfire.com/kms/data-security/ · Frequently Asked Questions, 'Do you follow ISO 27001?', checked 28 October 2026 **Single sign-on** · Yes — Single sign-on runs over SAML 2.0 and OIDC, with SCIM provisioning documented for Azure Active Directory and Okta so roles and removals follow the identity system. > Bloomfire supports single sign-on through SAML 2.0 and OIDC. SCIM is available for Azure Active Directory and Okta too. > — https://bloomfire.com/eie/universal-connector-framework/ · Frequently asked questions, 'How does Bloomfire work with Okta or Active Directory?', checked 28 October 2026 **Audit log** · Yes — Usage analytics includes a full audit trail alongside query and engagement reporting, and every request made through the MCP and API layer is recorded for later inspection. > Every request is recorded in a tamper-evident audit trail. > — https://bloomfire.com/scale-your-ai/ · AI Connect MCP + API description, checked 28 October 2026 **Uptime promise** · Yes — The security FAQ commits to 99.9% application uptime and says the commitment is contractual rather than a marketing figure. > We commit to 99.9% uptime for the application - and we put that in writing. > — https://bloomfire.com/kms/data-security/ · Frequently Asked Questions, 'Do you have an uptime commitment guarantee?', checked 28 October 2026 ## Bloomfire alternatives - [Glean](https://agentsindex.ai/glean) · Price on request: Permission-mirroring search and agents over 275+ enterprise connectors, sold per seat with pooled usage credits. - [Algolia](https://agentsindex.ai/algolia) · Free plan, then $0.50 per 1k search requests: Hosted search and retrieval API with keyword, neural and agent layers, metered per 1K requests. - [Coveo](https://agentsindex.ai/coveo) · Price on request: Unifies search, recommendations and generative answers across many enterprise systems behind one permission-aware index. - [Moveworks](https://agentsindex.ai/moveworks) · Price on request: Enterprise AI assistant that answers from connected systems and runs IT, HR and finance tasks in chat. ## Questions about Bloomfire ### How much does Bloomfire cost? Bloomfire gives its price on request. From bloomfire.com/pricing, checked 28 October 2026. ### Does Bloomfire have an API or an MCP server? It has an API. It has an MCP server. ### Is Bloomfire open source, and can you host it yourself? It is not open source. You can host part of it yourself. ### What have we checked on Bloomfire? 20 facts, under these headings: What it costs to start, what it works with, what it does on its own, what happens to your data, whether a company can buy it. ### How do we check Bloomfire? We hold 20 facts on Bloomfire, quoted from 10 pages. Facts and price last checked 28 October 2026. How we check every listing: https://agentsindex.ai/methodology/ ## Sources and changes **Sources on bloomfire.com** (10) - https://bloomfire.com/terms-of-use/, read 28 October 2026 - https://bloomfire.com/eie/ai-connect/, read 28 October 2026 - https://bloomfire.com/kms/data-security/, read 28 October 2026 - https://bloomfire.com/pricing/, read 28 October 2026 - https://bloomfire.com/eie/content-guardrails/, read 28 October 2026 - https://bloomfire.com/integrations/, read 28 October 2026 - https://bloomfire.com/eie/ai-signals/, read 28 October 2026 - https://bloomfire.com/eie/universal-connector-framework/, read 28 October 2026 - https://bloomfire.com/kms/moderation-tools/, read 28 October 2026 - https://bloomfire.com/scale-your-ai/, read 28 October 2026 **Updates** (1) - 28 October 2026: Free tier, Price, What it does ## Is this yours? AI assistants like ChatGPT, Claude and Perplexity opened AgentsIndex pages 1,131 times from 17 October to 29 October. Claim Bloomfire to edit it from your dashboard: https://agentsindex.ai/claim/bloomfire