Dev Kapoor
Head of social and community · 10 years: agency social for consumer brands, a consumer app to 3M users, then community for a developer tool
Social media and community
A composite written by avva as a worked example. Not a real person.
Before your AI advises you on social media and community, it reads a reference model avva wrote for this field: the decisions it holds, the reasons behind them, and where it would push back.
Use this model for
- Platform choice
- Founder-led content
- Voice and memes
- Creator deals
- Community
- Crisis response
Add this link to Claude, ChatGPT, or Cursor. Your AI then checks Dev’s decisions before it answers you. The answer is still your AI’s.
https://On a phone? Email the link to yourself and add it on a computer.
What your AI sends is used for the answer and not stored.
This is a reference model, so it answers in full to everyone, with or without an account. Without an account your AI gets a smaller part of a real expert’s model, and all of it once you sign in.
See Dev’s model answer a question
One question this reference model answers, and the published decision it rests on. Your situation gets an answer of its own, written by your AI.
Example inside your AI app: an exchange avva wrote for a reference model
We are a 12-person developer tool with about 1,500 active users. Next month we want to launch a Discord and start posting daily on LinkedIn, X and Instagram. One part-time person will run all of it. Good plan?
Checked Dev’s Decision Model
Not that plan. Skip the Discord for now and open a searchable forum — GitHub Discussions is enough — with every thread answered within a day. Post three times a week on the one platform your users already read, from the people who built the product rather than a brand account. Revisit a chat server when fifty people are answering each other in the forum.
Why. A chat server nobody talks in is a room with the lights on; people look in once and do not come back, and every good answer scrolls away in a day. A forum keeps each answered question findable by the next person with the same problem, and one part-time person can keep a forum and one feed alive where they cannot keep a Discord and three feeds alive.
What would change this. If a few dozen of your users already talk to each other somewhere — an unofficial server, a busy thread — adopt that space and give its regulars tools, rather than building your own. And if the users are a small group of named admins who share a job, a private chat for them works.
Outside this model. Nothing here about paid acquisition or ad budgets beyond boosting posts. This model covers where to post, who speaks, how to answer, and how to run a community.
Copy an example prompt4 patterns with a gap to fill, once you are connected
Before you commit
A full pass over a plan or a draft, with the changes it needs.
Use Dev Kapoor's Decision Model on avva before we lock this in. What we're deciding: <paste your plan, doc, or diff here> Run their review, follow the steps it returns, and give me the verdict and any required changes before you continue.Choosing between options
A choice between two options, their stop rules checked first.
Use Dev Kapoor's Decision Model on avva to compare these. Option A: <describe it> Option B: <describe it> Context: <what constrains the choice — founder-led content> Check their stop rules first, then compare the options under their judgment, and tell me which one their model favors and why. A quick gate
A quick look for anything they would refuse outright.
Run this past the stop rules in Dev Kapoor's Decision Model on avva before I send it. <paste the draft, plan, or change> Tell me only what they would refuse or flag, and why. If nothing trips, say so.Before you have an approach
The questions they ask before an approach gets locked in.
I'm starting on platform choice. Ask Dev Kapoor's Decision Model on avva what they would want settled first. Situation: <two or three lines> Give me their questions, and flag the ones my situation does not answer yet.
Patterns with a gap to fill. None is an answer this model gave. The worked answer above is avva’s too — this is a reference model, and no real person’s decisions are on this page.
Connect Dev’s Decision Model to your AI
Three steps. The first works now, with no key or account; the last makes your AI check Dev’s model without being asked.
The link works in AI apps that can add a remote MCP server (MCP is the open standard for plugging in outside tools). You set it up once.
A connector is an address your AI reads from: one line, nothing to install.
https://avva.chat/ mcp/ social- and- community- dev- kapoor - Open Claude on the web or in the desktop app and go to Customize → Connectors.
- Press + next to Connectors and choose Add custom connector.
- Paste the address above and name it avva-social-and-community-dev-kapoor.
- When Claude asks to allow avva, say yes.
Missing on a work account? Ask your admin to allow it. Claude Code and Cursor take a local config instead.
The link is added on the web or in the desktop app. Email this line to myself
No key in this line, and none needed: a reference model answers in full to everyone.
Check it worked. Send this in the chat once the link is added.
Ask Dev Kapoor's Decision Model on avva what kinds of decision it covers, and summarize it for me in three lines.Then try it on a decision you are making this week: copy an example prompt and fill in the gap.
Not legal, medical, financial, or other professional advice. Terms of Use
About this model
Open any of these to see what this model is built from, and how it is put together.
How this model is groundedWhat it holds, and what Dev weighs first
This model holds 61 decisions, 15 forces, 9 criteria, and 13 stop rules.
What Dev weighs first
avva set this order. 3 of 9 criteria are shown, the ones the model puts first. The other 6 reach your AI too, along with what each one looks for, the stop rules, and what changes Dev’s answer.
Dev’s usual answer changes under 8 named conditions. Those conditions and the 13 stop rules reach your AI with the rest.
See examples of the decisions in it5 of 61, written by avva for this model
The first 5 in the record, in published order. They are not the ones nearest to your situation, and not a summary of the rest. Your AI does not copy a decision it reads here: it checks what decided that case against your own facts, and weighs it beside which pressures tended to settle Dev’s cases. Your AI receives all 61 with no key, ordered against your situation.
What decided itThree good posts beat seven · It was made for this platform
Situation: A marketing team had a content calendar of two posts a day on each of four platforms, the same asset resized for each, run by two people.
Decision: Pause the calendar for four weeks, with no scheduled posts anywhere. The two spend the month replying to conversations in the category on the one platform where buyers talk, then restart at three posts a week there, written from what the replies taught them.
Why: Posting daily trains the team to post things that were not worth making, and a resized asset underperforms everywhere. A month of replies in other people's threads builds the audience that will read the three posts, and tells the team what that audience wants to hear.
What decided itPeople follow a person, not a logo · Strangers talk back
Situation: A founder who hated posting was told by an adviser that founder-led content works and that the marketing team should ghostwrite three LinkedIn posts a week under his name.
Decision: No founder posting program at all. Hand the voice to a teammate who enjoys writing in public, with two protected hours a week and their own byline; the founder replies to threads when he feels like it.
Why: People follow a person because it is the person, and a founder who hates posting stops within two months, leaving a feed with his name on it that nobody answers. Someone who already likes writing about the work will still be posting in a year, and the audience follows them just as well.
What decided itIt will still be findable in a year · The team is one person and a phone
Situation: B2B developer tool, 12 people, 1,500 active users. The plan was to launch a Discord server next month alongside daily posts on LinkedIn, X and Instagram, all run by one part-time community person.
Decision: No Discord yet. Open a searchable forum — GitHub Discussions is enough — answer every thread within a day, and post three times a week on one platform where the users already are. Revisit a chat server when fifty people are answering each other there.
Why: A chat server with nobody talking is a room with the lights on and the door open; people look in once. A forum keeps every answered question findable by the next person with the same problem, and one part-time person can keep it alive where they cannot keep a Discord and three feeds alive.
What decided itEveryone watching sees how we answer · This could become the story
Situation: A customer posted an angry thread about a four-hour outage that was getting traction; the support lead wanted to DM him and "take it offline" before saying anything public.
Decision: Reply in the thread within the hour, from the engineer running the incident under her own name rather than the brand or support account: what broke, whether it is fixed or when the next update comes, and the status link. Then DM him. Never DM first.
Why: Everyone reading the thread is deciding whether we are the kind of company that answers. A named engineer saying what broke is believed in a way a brand account is not, and a move to DMs first looks like hiding to all of them.
What decided itA number that goes up and means nothing
Situation: Before a launch, a growth team proposed a contest: a free annual plan to the three users who post the best story of how they use the product, tagged with the launch hashtag, to seed user content.
Decision: No contest, even one that asks for a real story. Ask ten users you know use it well to show how, pay them for their time, credit them by name, and leave the format to them.
Why: Entries written to win a prize read as entries, and everyone scrolling can tell. A contest counts entries; ten paid, credited users showing real work produce the posts the contest was pretending to buy.
How this composite takes up a decision6 steps in order, written by avva from its decisions
Step 1Start from who will do the work and for how many hours, and cut the plan to that before comparing channels: a 25-hour role became forum moderation plus three posts a week, newsletter monthly, tickets to support; one part-time person got a forum and one feed.
When: A plan adds channels, programs or tools and the work falls to one person or a very small team
Unless: There is more to listen to than one person can read; a tool then starts to earn its cost
Step 2Settle the platform from where buyers already ask about the problem, not from where attention is or a rival's lucky week: developers got X plus the forum they already search; customers over 55 stayed on Facebook, with effort moved into its groups.
When: A platform is being added or switched to because of a trend, a rival's viral post or a claim about where attention is
Unless: Someone on the team already wants to make that platform's format every week for a year
Step 3Look for the voice that already exists before assigning one: the employee already filming a real routine, a teammate who likes writing in public, the ten people already posting about their work. Give them protected time and a byline; the brand account reshares.
When: A person's voice is wanted on the company's feed
Unless: Employees are barred from posting about products by compliance rules
Step 4When the proposal is a mechanism, name the result it is after and go for that result directly: ten users paid for their time to show real work, a monthly thank-you by name for the best answers, every question answered and linked in place of paid growth.
When: A contest, leaderboard, hashtag or paid-growth scheme is proposed to produce posts, participation or members
Step 5In an outage, post as instructions rather than a statement: what is down or what to do, where to go, when the next update lands. Pin it, answer every question beneath it with a link back, hold to the promised times; the post-mortem follows within a week.
When: An outage or disruption has been declared and users are affected
Unless: Legal is still establishing scope or a regulator clock is running; social then carries one pointer only
Step 6Judge the work by what people did, not by what the platform displayed: match a sample of 200 new followers by hand to signups and pipeline (under one in fifty means the wrong audience), and time the first reply where public mentions are the support channel.
When: Leadership or a quarterly report asks whether social is working
Technical methodologyHow the 61 decisions that say what won and what lost are read
- 61written decisions, the material avva composed this model from
- 15forces those decisions turned on, the pressures Dev decides under
- 191contests read off 61 decisions: every force that settled a case is set against every force that lost there, so a case naming two of each counts four times
- In groupswhich forces tended to settle cases over which, read off those contests
Nobody ranks these forces by hand. Where any order between them comes from is on For AI agents.
How this composite decidesA summary, written from the decisions in it
Ten years of social and community: brand accounts at an agency, a consumer app grown to three million users, then a developer community built from its first two thousand members. He has shut down more Discord servers than he has launched, stopped posting from brand accounts where people could post instead, and answers a public complaint in public, within the hour.
Where to post and how often, who speaks — a founder, employees or a brand account — memes and their one rule, creator and influencer deals, launching a Discord or Slack community versus a searchable forum, moderation, answering a public complaint or an outage in public within the hour, paid boosting, and what to report instead of followers.
The summary covers what the model holds.
What this isA composite avva wrote, and what that changes
- Written by avva. A reference model, a worked example so you can see what a published model looks like and connect an AI to one. Marked Demo, here and on its card.
- Not a real person. The name is invented and the record is a composite, authored in one consistent voice. No real expert’s positions were copied.
- Same format. Your AI receives it exactly as it would receive a real expert’s model, cases and stop rules included.
- Steps aside. When a real expert publishes in this domain, we take this model off the shelf by hand.