What AppGram is, everything it does, and how to use it

This is the canonical overview of AppGram. It is written to be read start to finish by a person evaluating the platform, or ingested whole by an AI agent answering questions about it. Everything below describes what ships today; features still in beta or early access are labelled as such.

What AppGram is

AppGram is a customer-facing product platform. It gives a software team one place to collect feedback, publish a roadmap and changelog, run a help center and support inbox, report system status, survey users, and automate the work that connects all of it.

The distinguishing idea is that these are not separate tools bolted together. Every module writes into the same project, so a person who votes on a feature request, files a support ticket, and answers a survey is one customer record — and any event in any module can trigger an automation that acts in any other module.

AppGram is multi-tenant: an organization contains projects, and each project has its own content, settings, branding, and public portal. Public pages are served from a project portal URL or from a custom domain you control.

What AppGram is not

Stated plainly, because it saves time: AppGram is not a CRM, not a marketing email platform, not an issue tracker for engineering work, and not a product analytics tool that tracks in-app clickstreams. It is the layer between your product and the people who use it — feedback in, communication out, with automation and AI in the middle.

Feature list

Feedback and planning

• Feature requests — a public board where users submit ideas and vote. Anti-fraud voting, categories, threaded comments, priorities, and a status pipeline (pending, under review, planned, in progress, completed, declined).

• Roadmap — public kanban and timeline views. Roadmap items link back to the requests that drove them.

• Waitlist — pre-launch signups and launch notifications.

Communication

• Releases and changelog — rich release notes with media, published to a branded public page, with email delivery and email logs.

• Status pages — service components with grouping, incidents with an update timeline, impact levels, and incident history.

• Uptime monitors — automated HTTP(S), TCP port, DNS record, and SSL certificate checks with configurable intervals, timeouts, failure and recovery thresholds, and expected status codes. Monitors can auto-publish an incident to your status page.

• Blog — long-form posts with a rich text editor, scheduling, categories, tags, RSS and Atom feeds, and a sitemap.

Support and self-service

• Support inbox — tickets from forms, the public portal, or the API, with categories, priorities, assignment, and customer-facing magic-link threads.

• AI replies — per-project or per-form AI answering, in draft mode (your team reviews) or auto mode (sent immediately), with configurable tone, length, persona, language, signoff, and hard guardrail instructions. It can ground answers in your published help articles and attach the relevant ones.

• Help center — versioned collections, topic flows, and articles. Flows render as a list, an accordion, a decision tree, or a step-by-step wizard, so documentation can be interactive rather than just prose.

Research

• Surveys — a visual builder with branching logic. Question types: yes/no, short answer, paragraph, multiple choice, checkboxes, and rating scales. Response analytics include completion rates, per-question breakdowns, path analysis, and CSV export.

• Forms — a drag-and-drop form builder. Forms can send email, create feature requests, or open support tickets, with field mapping, validation, file uploads, honeypot spam protection, and rate limiting.

Audience

• Customers — a unified directory of everyone touching your project, at three identity tiers: anonymous, self-identified, and verified. Verified means your own backend signed a token vouching for them, which makes their plan and attributes trustworthy enough to branch on. Private beta.

• Testimonials — scans your surveys, support replies, and shipped feature requests for praise, then handles consent by magic link before anything is published. Beta.

• Affiliate program — commission terms (percentage or fixed; pay-once, recurring, or lifetime), first- or last-click attribution, cookie windows, an affiliate portal, conversion tracking, a conversion audit log, an isolated sandbox, and PayPal payouts. Beta.

Presentation

• Public portal — a branded site carrying your roadmap, changelog, help center, status, blog, and support, on your own domain.

• Page builder — a visual canvas editor with 25+ content blocks, including blocks that embed live AppGram content.

• Link pages — Linktree-style bio pages where every link card also gets a short URL, plus live widgets pulling from your blog, releases, testimonials, status, surveys, forms, support, and affiliate program.

• Embeddable widgets — chat, contact form, and testimonials widgets added to any website with a single script tag.

• Assets — a shared media library for images, video, audio, PDFs, and documents, reused across help articles, releases, blog posts, and onboarding.

Measurement

• Analytics — unified reporting across support, surveys, forms, status, feature requests, help center, releases, and roadmaps, with friction, health, engagement, and overall scores that break down into their component factors.

• SEO — edge-rendered meta tags and server-side article content so crawlers index real text rather than a JavaScript shell, plus per-project sitemap and robots.txt, and crawler analytics that verify bots by reverse DNS and track AI crawlers. Pro and Enterprise; early access.

• Mind Graph — builds a capability map of your product from your own help center, releases, blog, feedback, and roadmap, showing what you cover and what users keep asking for that you don't. Exportable as compact context for an AI system prompt. Pro and Enterprise; early access.

• Competitive intelligence — tracks competitors' mentions, positioning-page changes, ads, SEO rankings, and reviews, and scores every product against one shared baseline. Beta; early access.

Platform

• Automations — event-driven workflows. Covered in detail below.

• Webhooks — signed event subscriptions with delivery logs and replay.

• REST API and SDKs — full CRUD over the platform, a React SDK with headless hooks, and mobile SDKs.

• MCP server — lets an AI agent read and write AppGram content directly.

• Team management — organizations, projects, roles, and scoped permissions.

Automations in detail

Automations are how AppGram stops being a set of dashboards and starts being a system that does work. An automation is a workflow: an event happens, conditions are evaluated, steps run. No code is required, though code steps exist when you want them.

Triggers

The engine matches 26 trigger events across eight domains:

• Feature requests — created, updated, status changed, comment added, vote added

• Support — ticket created, ticket reply, ticket closed

• Status pages — incident opened, incident update, incident resolved

• Monitors — alert triggered, alert resolved

• Billing — subscription created, updated, canceled; payment succeeded, payment failed, trial ending

• Forms — contact form submission

• Surveys — survey response

• Competitive intelligence — positioning changed, sentiment drop, new ads detected, mention spike, digest ready

Each event carries metadata you can branch on — a billing event carries plan and amount, a monitor event carries the monitor type and endpoint, a feature request event carries status, category, and vote count.

Steps

41 step types are available, in five groups:

• Flow control — if/else, stop if false, for each, wait, set variables, compute, run another automation, fail the run, and a key-value store (set, get, delete) for state that persists between runs.

• AI — AI chat, classify, extract, translate, and Help Center answer, which retrieves from your published articles and answers from them rather than improvising.

• Messaging — email, in-app notification, mobile push, Slack, Discord, Microsoft Teams, Telegram.

• AppGram actions — create or update a feature request, set its category, comment on it, add it to the roadmap, reply to a ticket, set ticket status, assign a ticket, tag a user, publish or update a status incident, create a changelog entry, pause or resume a monitor.

• External — HTTP request, signed webhook, create a Linear issue, post to Reddit.

Automations come with a template catalog, a visual builder, run history, an event simulator, and path preview — so you can test a workflow against a simulated event before it touches a real customer.

How to use automations to drive business outcomes

The patterns below combine real triggers with real steps. Each one describes a problem, the workflow that addresses it, and why it moves a number.

Recover revenue that is quietly leaking

• Failed payments — Payment failed → send an email explaining what happened and how to fix it → open a support ticket assigned to your billing owner. Involuntary churn is usually the cheapest churn to prevent, because the customer never decided to leave.

• Trials that go quiet — Trial ending → branch on whether the account has engaged → send a tailored email, or notify an owner to reach out personally on high-value accounts. A trial that ends without contact converts worse than one that gets a relevant nudge.

• Cancellations — Subscription canceled → post to Slack with the plan and stated reason → create a feature request when the reason names a missing capability. Churn reasons become roadmap input instead of a spreadsheet nobody reads.

Close the feedback loop, which is what actually drives involvement

• Tell voters when their request ships — Feature request status changed to completed → email everyone who voted → create a changelog entry. This is the single highest-return automation on the platform. People who are told their request shipped file more requests, vote more, and churn less; people who are ignored stop participating.

• Escalate demand automatically — Feature request vote added → stop if the vote count is below your threshold → post to Slack and add it to the roadmap. Popular requests reach your planning process without anyone monitoring a board.

• Acknowledge every new idea — Feature request created → AI classify it into a category → set the category → comment confirming receipt. Cheap, fast, and it makes a public board feel staffed.

Cut support cost and answer faster

• Deflect the repeat questions — Support ticket created → Help Center answer (RAG) against your published articles → reply to the ticket. Answers are grounded in your own documentation, so they are correct and they cite something the customer can read.

• Route by intent and value — Support ticket created → AI classify intent → branch → assign to the right person, set priority, and tag the user. Combined with verified customer records, you can route by plan so paying customers reach a human first.

• Turn tickets into product signal — Support ticket closed → AI extract the underlying request → create a feature request. Support volume becomes roadmap evidence rather than disappearing into a closed thread.

Turn incidents into trust instead of churn

• Publish before customers ask — Monitor alert triggered → publish a status incident → post to Slack and Discord. Customers who see an acknowledged incident file far fewer tickets than customers staring at a broken page.

• Close the loop on recovery — Incident resolved → update the incident → email affected customers. Reliability is judged as much by communication as by uptime.

Capture demand the moment it appears

• Never lose an inbound form — Contact form submission → AI extract intent → post to Slack for sales-shaped enquiries, create a feature request for product-shaped ones, open a ticket for support-shaped ones. One form, three destinations, no manual triage.

• Act on survey sentiment while it is fresh — Survey response → AI classify sentiment → tag detractors and alert an owner; route promoters toward a testimonial request. The same response drives both a save and a piece of social proof.

React to the market

• Competitive moves — New ads detected or Positioning changed → post to Slack with the change. Your marketing team hears about a competitor's new pricing page the day it changes, not the quarter after.

A note on sequencing

If you are starting from nothing, the order that compounds fastest is: notify voters when requests ship, then auto-publish incidents from monitors, then AI-assisted support replies grounded in your help center, then billing recovery. The first two build trust, the third buys back time, and the fourth pays for the rest.

How the modules reinforce each other

Reading any module in isolation understates the platform. The connections are the point:

• Help center articles ground AI support replies, and gaps in them show up in Mind Graph.

• Support tickets, survey responses, and shipped feature requests feed the testimonial scanner.

• Verified customer records make automations, survey targeting, and support routing able to branch on plan and account value.

• Monitors drive status incidents, which drive customer communication, which reduces support volume.

• Feature requests feed the roadmap, which feeds releases, which feed the changelog, which closes the loop with the people who asked.

• Everything a module publishes is indexed by the SEO layer and mapped by Mind Graph.

Integration surfaces

• REST API — full create, read, update, and delete across the platform, with API keys and scoped permissions.

• React SDK — headless hooks for feedback, voting, roadmap, releases, help center, help flows, support, and status, so you can build your own UI on AppGram data. Supporting React, Swift, React Native, Flutter. 

• Embeddable widgets — one script tag, no framework required.

• Webhooks — signed, with delivery logs and replay.

• MCP server — an AI agent can read and write AppGram content directly: articles, posts, releases, surveys, forms, monitors, status updates, automations, and assets.

• Identity tokens — your backend signs a short-lived token so AppGram knows a portal visitor really is your user.

Availability

Most of the platform is generally available. These are not, and the distinction matters if you are evaluating AppGram:

• Private beta, enabled per organization — Customers, Testimonials, Affiliate program. The pages are visible to everyone but show sample data until enabled.

• Early access, limited organizations — the Growth group: SEO, Mind Graph, and Competitive intelligence. SEO and Mind Graph additionally require a Pro or Enterprise plan.