Transparency for the AI front office: what we promise, what we have proven, what is in progress, and what we will not claim.
AI companies make a lot of claims. We believe you deserve to know which claims are backed by evidence, which are in progress, and which we explicitly don't make. This is our public accountability ledger.
Bob only answers from content you've approved—including your business profile when you've completed it—so answers stay grounded in what you chose to publish. Every response can be traced to its source.
When Bob can't ground an answer in your content, Bob says so and offers to escalate.
Complete logs of what Bob said, what sources were used, and when conversations were escalated.
Bob only takes actions you've explicitly enabled (booking, quoting, etc.).
Bob keeps improving from your training and from how we ship the product. We run automated checks so learning-related updates don't casually break answers that already worked. Stretching every improvement across every edge case—with no surprises—is engineering work we treat as never “finished,” which is why this promise stays in progress.
Same Bob and the same approved knowledge across the channels we turn on: website chat is live today; voice on your site is in beta; your main business phone line plus SMS and WhatsApp are in active rollout (phone target Q2 2026—Professional is the tier we’re aligning with those channels when they go live).
Bob can surface suggestions, gaps, and candidates for what to remember next. When your account uses our review flow—we call it Bob's Review in the Customer Portal—lasting memory does not change behind your back. You approve or decline there. Enablement is deliberate: your team or ours can turn this on gradually (pilots and staged rollout), not as a surprise default for every tenant overnight.
Updated May 2026 — promises reviewed quarterly.
Join the Waitlist