Operating principles · 6 01
AI that works
An AI feature that looks good in a demo but doesn't save the team time is worse than no AI. Every capability is measured against a before/after workflow metric. We kill AI features that fail that test.
Operating principles · 6
How we
operate.
Six operating principles that shape how DialPhone builds product, runs support, prices, and treats customer data. Written down so we can hold each other to them.
Operating principles are easy to write and hard to live by. The six below were not drafted as marketing — they were written after specific disagreements where the team needed a tie-breaker that did not depend on whoever was loudest in the room. Each one cost something real to commit to: shipping fewer AI features, paying for more redundancy than the SLA strictly required, refusing pricing exceptions that would have closed deals faster, publishing post-mortems that named our own mistakes.
"AI that works" is a constraint, not an aspiration. We have killed AI features in beta when the before/after metric did not move; specifically, an early AI auto-response feature that scored well on user surveys but failed to improve reply rates was shut down. "Uptime is trust" budget shows up in headcount: more engineers on infrastructure than on most feature teams. "Privacy is the product" shows up in the privacy commitments page and the subprocessor registry being public.
"Transparent pricing" is the easiest one to verify externally. Every tier on the main pricing page, every CCaaS tier on the contact center pricing page, and every AI add-on on the AI add-on page is published. The open VoIP Pricing Research 2026 report exists in part to make it impossible for any vendor in the market — including ourselves — to obscure published-vs-realized cost differences.
Operating principles
-
-
02
Uptime is trust
99.999% SLA. A phone system that drops a customer call costs the business more than the plan's price. We budget reliability as a first-class concern, more engineers on infrastructure than on most feature teams.
-
03
Privacy is the product
Customer conversations belong to the customer. We don't train shared AI models on customer data. We publish subprocessors. We sign HIPAA BAAs at no surcharge. Privacy is a property of the platform, not a marketing claim.
-
04
Transparent pricing
Every tier published, UCaaS, contact center, AI add-ons. No "call for a quote" until enterprise-specific custom work. Procurement teams should spend time evaluating fit, not extracting numbers.
-
05
Candor with customers
If we lose on a specific feature against a competitor, we say so on the compare page. If we hit an incident, we publish the post-mortem. Long-term trust beats short-term positioning.
-
06
Small teams, big outcomes
We hire generalists with deep expertise, not narrow specialists. A typical DialPhone product team is 4-8 people shipping features that larger teams at competitors take quarters to deliver.
Related