The Dashboard That Learned to Answer Back
A community-management dashboard that turned itself into a support AI - how two ex-Zomato builders taught software to answer the questions people keep asking.
In 2021, a small team in Bangalore built the kind of software that felt inevitable for its moment. Online communities were exploding - every product had a Slack, every creator had a Discord - and nobody could keep track of any of it. Threado offered a fix: one dashboard that pulled together Slack, Discord, Discourse, Twitter and GitHub, showed you who was going quiet, and flagged the questions nobody had answered. It called itself the command center for your online community.
The insight underneath it was sharper than the pitch. Pramod Rao had spent about eight and a half years at Zomato building user growth and community before he co-founded Threado with the engineer Abhishek Nalin. He had watched the same thing happen over and over, even to communities that looked healthy. "Even highly engaged communities," as he put it, "tend to zero with time in terms of activity." Threado was built to fight that slow decay - to give community managers, in his words, "superpowers."
Then the team noticed something in their own data. The hardest part of running a community was not the engagement graph. It was the questions. The same ones, asked again and again, scattered across channels, half of them never getting a reply. And that is where the story turns.
01 / THE PIVOTOne product, two lives
By 2024, Threado was not really a community dashboard anymore. It had become Threado AI: a system that builds custom AI agents trained on a company's own knowledge - help docs, past tickets, internal wikis, old conversations - and puts them to work answering questions. The community-tracking heritage did not vanish so much as get repurposed. The team had always been in the business of unanswered questions. Now they were teaching software to answer them.
The pivot is the interesting part, and not because pivots are rare. Most startups from the 2021 community-tech wave either stalled or shut down when the hype cooled. Threado read its own users, found the deeper problem hiding inside the surface one, and moved the entire product toward it. Same founders, same core observation, new shape.
02 / THE PRODUCTWhat it actually does
Strip away the category labels and Threado does one thing: it makes a company's scattered knowledge answer for itself. A support team feeds the agent its documentation, its helpdesk history, its Notion and Confluence pages. The agent learns the company's own words. Then it sits where the team already works - inside Slack, inside Microsoft Teams, or as a Chrome extension riding along in the browser - and replies when someone asks.
Two products carry that idea. AI Agents, launched in mid-2024, handle the customer-facing side: instant answers, deployed in minutes, grounded in the company's material rather than the open internet. Support OS, released earlier that year, works the other direction - it sits beside human agents inside ticketing tools, drafting replies and surfacing the right knowledge in real time so a person spends less of the day hunting for an answer they half-remember.
The design choice that matters most is the least flashy one. Threado did not build another tab to open. It put the AI inside the tools nobody ever closes. For a support team, the difference between a new app and a message in Slack is the difference between a tool you adopt and a tool you actually use.
There is a second, quieter bet in how the agent is trained. Plenty of tools will bolt a general-purpose model onto a chat box and hope it sounds right. Threado's version starts from the opposite end: the agent is grounded in the company's own material first, so its answers carry the company's own facts and phrasing rather than a plausible-sounding guess. For anyone who has watched a generic bot confidently invent a refund policy, that distinction is the whole game. It is also why the security posture matters - reviewers note the option to run custom models when a company's data rules demand it.
Every repeated question the agent answers is a ticket a human never touches. That share - illustrative here - is the whole business case for support AI, and the number Threado is built to move.
03 / THE CUSTOMERWho it is for
Threado's users are the people drowning in the same three problems: a knowledge base nobody can navigate, a ticket queue that never empties, and a set of questions asked so often they have become background noise. That describes customer-support teams, community managers, and internal-operations people at startups and mid-market software companies - teams big enough to feel the repetition, small enough that hiring their way out of it is not an option.
The early reviews on G2, Capterra and Product Hunt come from exactly that crowd, and they tend to praise two things: how fast the team shipped features, and how quickly the agent could be stood up. Setup measured in minutes, not a rollout measured in quarters.
04 / THE MONEYWho backed it, and why
In 2022 Threado raised about $3.1 million in a seed round led by Vertex Ventures, with Gemba Capital alongside. The angel list is the tell. It included Zomato's founder Deepinder Goyal and co-founder Pankaj Chaddah - Rao's old world - plus Krish Subramanian and Rajaraman Santhanam, the founders of Chargebee. When operators who have built Indian software at scale write personal checks, they are usually buying the founder as much as the pitch.
The model is straightforward B2B SaaS: subscriptions for the agents and the support tooling, priced by usage, seats and integrations, sold to the teams who feel the pain of repeated questions most acutely.
The expertise behind it is worth naming, because it explains the shape of the product. Rao is not a support-software lifer; he is a growth and community operator who spent nearly a decade watching how people actually behave at scale inside a consumer giant. Nalin is the builder who translated that behavioral read into software. That pairing - someone who has felt the problem in his bones, and someone who can ship the fix - tends to produce products that solve the real thing rather than the demo of the thing. Threado's willingness to abandon a working dashboard for a harder, more valuable problem is the kind of move that comes from operators, not tourists.
05 / THE FIELDWhere it sits
Support AI is now a crowded room. Intercom has Fin. Zendesk has its own AI layer. Freshworks is pushing hard, and a wave of newer names - Decagon, Forethought - are chasing the same promise of automated, accurate answers. Threado's position was the pragmatist's one: not the biggest, but grounded in a company's own knowledge and living inside the tools teams already run, rather than asking them to migrate.
A note on the present tense. Some third-party trackers list Threado Inc. as having wound down around March 2025. That status is reported second-hand and is best treated as approximate rather than confirmed. What is not in doubt is the shape of what the company built, and the clarity of the idea behind it - which is the part worth studying regardless of where the corporate entity ends up.
06 / THE LESSONWhat you can take from it
You do not need to buy anything to use Threado's best idea. Pull your last few thousand support tickets and count how often the same twenty questions recur. Those are not tickets. They are an FAQ your customers could not find, and every one of them is a small tax on your team's day. Threado turned that observation into a company. Any team can turn it into an afternoon of cleanup.
The other lesson is about listening to the boring parts of your own data. Threado did not pivot on a whim or a headline. It pivoted because the dull, repetitive thing its users kept doing - answering the same question - was the actual product hiding inside the exciting one. The founders had the discipline to follow the boring signal.
What began as a way to keep communities from fading to zero became a way to keep support teams from repeating themselves. Different surface, same underlying belief: that the question people keep asking is a problem worth solving properly, once, so nobody has to answer it by hand again.