Multi-model chat platform BrainChat discloses runtime spend caps and compliance certification, rare among tracked AI vendors

BrainChat, a browser-based workspace that gives teams a single interface to query multiple large language models, was added to AI Dispatch's index this week with disclosed evidence on two of the five dimensions in the index's trust-gap rubric — compliance and runtime governance — where the majority of tracked vendors publish nothing at all.

The company describes itself as "The AI workspace for modern teams," according to its site, offering "access to OpenAI, Anthropic, Google and DeepSeek" and letting users "Compare side-by-side and pay one bill." Teams supply their own API credentials: "Bring your own keys for OpenAI, Anthropic, Gemini, DeepSeek & Mistral," the site states.

On compliance, BrainChat's site states it is "SOC 2 Type II & GDPR compliant." On runtime governance — whether an AI system's scope, spend or authority are enforced while it operates, rather than described only as policy — the same page states its "Admin dashboard: per-seat budgets, SSO, model allow-lists," indicating administrators can cap spend per seat and restrict which models a team can access.

AI Dispatch's review found no certificate, audit scope or attestation report accompanying the SOC 2 claim, and no description of whether the budget and allow-list controls are enforced automatically or are configurable settings an administrator could leave unset. The index has not yet assigned BrainChat numeric trust-gap scores.

On the rubric's other three dimensions, BrainChat's disclosures match the pattern found across most vendors AI Dispatch tracks: nothing is published about audit-trail immutability, about payment tied to a verified outcome rather than seat-based subscription, or about insurance or indemnification. Fields covering safety evaluation, incident reporting, red-teaming and third-party audits are also marked "no public information" in AI Dispatch's records.

Of the fields AI Dispatch tracks for BrainChat, evidence is present for 18 and absent for 37, according to the index's per-field citations, all drawn from the company's own site.

AI Dispatch's full rubric and per-field sourcing are published at aidispatch.news.