Not every change deserves an immediate email. At Tallyfy, we designed three notification frequencies: Electric, Mindful, and Chilled to give watchers control over their own inbox load. The tension between real-time updates and respectful batching shaped everything about how we aggregate and deliver watching notifications.
At Tallyfy, the 2017 debate over process-level managers versus step-level owners shaped every assignment feature built since, from bulk permissions to group-based ownership.
Why Tallyfy designed workflow assignment around predefined groups instead of individuals. Making all users belong to all groups by default forces role-based thinking and eliminates template updates when employees change roles.
The seven-year evolution from customer request to elegant two-rule solution at Tallyfy. Why checkboxes beat flowcharts for handling complex approval rejections and re-work scenarios in workflow automation.
The moment an anonymous visitor becomes a tracked participant at Tallyfy. How we designed public kickoff forms that let anyone launch a workflow, and why email validation turns random submissions into real process runs.
Nobody reads their 47th task reminder. The Tallyfy engineering team learned this the hard way building 77 email templates across 6 locales. The daily digest became our answer, an activity-stream approach that pulls people in instead of pushing them away. Here is the engineering story behind reminder emails that actually get opened.
Building Tallyfy Sherlock, an if-this-then-that rules engine for workflow automation, from April 2018. The internal debates, the disagreements on scope, and why testing rules before deployment became the defining feature.
Why Tallyfy built custom SAML 2.0 authentication instead of paying Okta or Auth0. The real cost of enterprise SSO, certificate management, SCIM provisioning for organizations with 5000-plus users, and a private key exposure vulnerability in API responses.
Most companies already have SOPs in Word or PDF format. Tallyfy built document-to-workflow parsing from 2017 to present, evolving from flowchart annotation ideas to AI-powered conversion that processes complex documents in 24 seconds or more.
Swimlane diagrams capture three dimensions most workflow tools miss: what needs doing, who does it, and when. A typical employee review takes 143 minutes of work spread across 17 business days and four departments. Tallyfy tracks all three at the template level so handoffs stay visible.
The biggest problem in workflow software was never technical. It was getting team agreement on who does what. This 2017 insight at Tallyfy shaped eight years of product decisions.
Templates and processes live in separate folders at Tallyfy because they represent different modes of work. The blueprint versus instance mental model, influenced by tag-based models like Gmail, took years to communicate effectively.
When your shipping process differs slightly by country, most teams clone their template fifteen times. This creates a maintenance nightmare. At Tallyfy, we designed a better approach: define variations from Standard and visualize the difference between one variation and another.
The Tallyfy permission matrix for watchers turned out simpler than expected. If you can edit an object, you can edit its watchers. That single rule drove most of the design decisions behind three watch ownership types.
Favorites in a business context does not really mean anything. At Tallyfy, we redesigned the entire notification system around what Jakob Nielsen called the 1% rule - 90% of users observe rather than create. The familiar star icon stayed but everything underneath changed to support three notification frequencies and process-level watching.
At Tallyfy, we started building a separate EPIC for webhook subscriptions, letting users subscribe to specific events via API endpoints. Then we realized it overlapped with the watching system. The notification target became just another channel: email, webhook, or chat.
How Tallyfy engineered deadline calculations that respect business hours, weekends, and timezones. The 2-hour default disaster, the Friday 4:30pm problem, and why user-level working hours had to wait.