Back to vendors
B

Brainfish

Also known as: Brainfish Pty Ltd, Ask Brainfish, Brainfish AI

Visit site
Entry priceContact for pricingFull pricing detail

AI customer support agent that answers only from your approved knowledge sources, cites every source, and resolves questions across chat, email, in product, and Slack.

Brainfish is an artificial intelligence customer support platform positioned as AI support agents designed for business to business teams. Founded in Sydney in 2023 and now operating from Sydney and San Francisco, the company raised a six point four million dollar pre Series A led by Prosus Ventures in 2025 and is run by co founder and chief executive Daniel Kimber. It began as an instant answer engine and has since repositioned around grounded agents that resolve customer questions across a company's website, product, help center, and messaging tools rather than simply deflecting them.

The foundation of the product is a self updating knowledge layer. Brainfish ingests a team's documentation, tickets, recorded calls, and conversations from tools like Notion, Confluence, Google Drive, Loom, Zendesk, Salesforce Knowledge, Help Scout, Helpjuice, Mintlify and ReadMe, along with OpenAPI specs, websites and any HTTP source, then routes the exact slice each agent needs at the moment of a question.

Its defining discipline is that it answers only from approved sources and cites the exact documents behind every response, with a confidence score attached, so there is no fabrication or paraphrase. It auto detects content drift as a product ships, flags stale articles and can rewrite them, and drafts new articles from calls and Slack threads.

The same knowledge layer powers every surface a team uses, so answers stay consistent across an embedded chat widget, an AI powered help center, a contextual agent inside the product, and the same agent inside Slack and Teams. Answers are personalized with live account, product, and user context, and the agent sends replies directly for cases a team trusts while escalating the rest to a human with full conversation context.

Brainfish also drafts ticket replies inside the helpdesk a team already runs, positioning itself as the knowledge layer underneath tools like Zendesk and Intercom rather than a replacement for them. An AI Actions add on takes it past answering into automating workflows and advanced support tasks through tool calls and custom API integrations, configured without code.

For measurement, Brainfish tracks resolution rates, deflection, and customer satisfaction, surfaces the questions its knowledge layer is failing to answer, and recommends improvements to specific agent responses, closing the loop between insight and content.

On security it publishes a trust center and carries SOC 2 Type II audited annually, ISO/IEC 27001:2022 certification, and General Data Protection Regulation compliance, with single sign on through Okta or Azure AD, customer managed encryption keys, private tenants, and a choice of United States, European Union, or Australia hosting that stays in region for both storage and inference. It never trains its models on customer data.

It also ships a Model Context Protocol server so engineering teams can plug Brainfish into their own artificial intelligence surfaces, though it publishes no API reference or SDK for its own platform. It does not publish prices. The plan structure and quotas are on the pricing page, but every number comes through sales.

Vendor details

Canonical URL

https://brainfishai.com

Category

Customer support agent

Company status

independent

Use cases & customers

In practice

A software user gets stuck on a complex feature at midnight. Rather than file a ticket, they ask Brainfish inside the product and get an instant answer from approved docs, with the exact source cited.

A support team is buried in repetitive how to questions. Brainfish deflects the routine ones with cited self service answers across the help center and in product, freeing agents to focus on the genuinely complex cases that need human expertise.

An internal team keeps asking the same questions in Slack. Brainfish answers them directly in Slack from the company's own knowledge, so the platform support channel stops getting clogged with repeat requests.

Agentic Index coverage score

7.5 / 14 capabilities · 54%

Integrations & Tool Calling Full

The agent reads from and writes to the tools a team already runs, so it acts as well as retrieves. The named tools span helpdesk (Zendesk, Intercom, Help Scout), CRM (Salesforce, HubSpot), messaging (Slack, Microsoft Teams), documentation (Notion, Confluence, Mintlify, ReadMe), storage (Google Drive), mail (Gmail), issue tracking (Linear, Jira), revenue intelligence (Gong) and video (Loom). It also works with OpenAPI specs and any HTTP source.

Sourcebrainfishai.comread 2026-09-05

Workflow Orchestration Partial

Beyond retrieving, answering and escalating, an AI Actions add on automates workflows and handles advanced support tasks through custom API integrations, tool calls, MCP and callbacks. It is configured with no code or low code and sold in blocks of five actions. A Workflows surface is marked as a new capability, but Brainfish does not say what it does. Brainfish describes no multi agent execution, branching runtime or handoff from one agent to another.

Sourcebrainfishai.com/pricingread 2026-09-05

Knowledge Grounding & RAG Full

Content flows into a single maintained knowledge layer from Notion, Confluence, Google Drive, Loom, Zendesk, Salesforce Knowledge, Help Scout, Helpjuice, Mintlify, ReadMe, OpenAPI specs, websites and any HTTP source. The layer is kept current by flagging stale articles and drafting new ones from calls and Slack threads, so it is maintained over time and not assembled for each question. At the moment of a question it routes the slice each agent needs. Every answer is drawn only from approved sources and cites the exact documents behind it with a confidence score.

Sourcebrainfishai.com/pricingread 2026-09-05

Human Oversight & Guardrails Partial

Grounding limits what the agent can say. It answers only from approved sources and cites them, but nothing is reviewed before it replies. Autonomy is set by a trust threshold. The agent sends directly for cases a team trusts and escalates the rest to a person with full conversation context, under configurable rules. Escalation hands the case over and does not hold a reply for sign off. Where a person does approve, it is a draft reply inside Zendesk or Intercom, and that approval step belongs to the helpdesk, not to Brainfish, which has no approval surface of its own.

Sourcebrainfishai.comread 2026-09-05

Security, Identity & Governance Full

Brainfish holds SOC 2 Type II, audited annually by an independent third party, and ISO/IEC 27001:2022 certification, and runs a Trust and Safety center at trust.brainfi.sh. It complies with GDPR, with right to delete built in, and commits to not training models on customer data. Single sign on supports Okta and Azure AD as identity providers.

Access controls also include private tenants and, for enterprise, customer managed encryption keys. Sensitive fields are obfuscated before any snapshot, and snapshots are processed then discarded by default. Brainfish does not say how a customer obtains the SOC 2 report.

Sourcetrust.brainfi.shread 2026-09-05

Observability & Auditability Partial

Reports cover resolution rate, deflection and customer satisfaction, and the product surfaces the questions its knowledge layer is failing to answer. Every answer cites the exact source documents behind it with a confidence score, so a reader can see which document produced a given reply. Brainfish describes no step level decision trace, tool call log or configuration audit trail.

Sourcebrainfishai.comread 2026-09-05

Memory & State Persistence Not documented

Answers are personalized with live account, product and user context supplied at the moment of the question. Sensitive fields are obfuscated before any snapshot is taken, and snapshots are processed and discarded by default, so that state is deliberately not kept. The one persistent store is the knowledge layer, which holds the team's content. Brainfish describes no memory for its agents.

Sourcebrainfishai.com/pricingread 2026-09-05

Deployment & Data Residency Full

Hosting runs in the United States, European Union and Australia, and the customer chooses the region. Data stays in region and does not leave it for inference or storage. EU hosting carries no quota, SLA or support difference from the US region, and it is available on Platform and above. Private tenants and customer managed encryption keys are offered on top. There is no self hosted or on premises build.

Sourcebrainfishai.com/pricingread 2026-09-05

Prebuilt Agents, Templates & Packs Not documented

The product ships as a single configurable support agent pointed at the customer's own knowledge. The onboarding pitch is a working agent built against the customer's own content, not one chosen from a set. Brainfish names no prebuilt agent, template gallery, cloneable starter or marketplace of packaged assets.

Sourcebrainfishai.comread 2026-09-05

Triggers & Channel Coverage Full

Intercom live conversations and Zendesk tickets flow in from the helpdesk a team already runs, so queues bring work to the agent as well as channels. Users can ask in Slack, get answers in Microsoft Teams, or use the embedded chat widget, the AI help center, an in product agent and email. Proactive prompts surface before a user asks. Brainfish names no scheduled or cron style trigger.

Sourcebrainfishai.comread 2026-09-05

Model Flexibility & Routing Not documented

Brainfish names no model provider and describes no model selection, routing or bring your own model control for customers. It says only that models are not trained on customer data.

Sourcebrainfishai.comread 2026-09-05

APIs, SDKs & MCP Extensibility Partial

A native Model Context Protocol server lets engineering teams plug Brainfish into their own AI surfaces, with Claude named as a supported client. The AI Actions add on is sold on APIs, tool calls, MCP and callbacks, and those APIs are the customer's own systems, which the agent calls. Brainfish publishes no public REST API reference or SDK for its own platform. Its developer resources are a changelog and a status page. ReadMe, Mintlify and OpenAPI specs are knowledge sources Brainfish ingests from, not its own developer documentation.

Sourcebrainfishai.com/pricingread 2026-09-05

Testing, Debugging & Optimization Partial

Live performance feeds back into content. The product surfaces the questions its knowledge layer is failing to answer and recommends improvements to specific agent responses. This optimization loop runs after deployment and is measured against resolution and deflection. What it improves is the knowledge, not the agent's configuration. Brainfish describes no test suite, conversation simulation, scenario set or scored evaluation of the agent itself.

Sourcebrainfishai.comread 2026-09-05

Browser & Computer Use Not documented

The agent reaches other systems programmatically, reading from and writing to the tools a team already runs through named integrations. Brainfish describes no browser control, hosted session, virtual desktop or computer use.

Sourcebrainfishai.comread 2026-09-05

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Pricing

Contact for pricing

Free / trial

No free tier or trial published; demo on request

Agentic Index verified 2026-09-05

Alternatives to Brainfish

The closest documented capability profiles to Brainfish among customer support agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

  • Freshworks10.0 / 14Adds documented Prebuilt Agents, Templates & Packs
  • Genesys10.0 / 14Adds documented Memory & State Persistence and Prebuilt Agents, Templates & Packs
  • BlueTweak9.5 / 14Adds documented Memory & State Persistence and Prebuilt Agents, Templates & Packs
  • Fini8.5 / 14Adds documented Prebuilt Agents, Templates & PacksBrainfish vs Fini →
  • Help Scout8.5 / 14Adds documented Prebuilt Agents, Templates & Packs
  • Maven AGI8.5 / 14Adds documented Prebuilt Agents, Templates & Packs

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.