Senior Solutions Engineer @hops_ai & @zencoderai | AI Agents
India
Joined October 2014
- Tweets2.6K
- Following1.3K
- Followers1.5K
- Likes13.6K
One support escalation can involve a ticket, a Slack thread, a Jira issue, a Sentry trace, and a doc. Switching tabs costs less than re-explaining the customer's problem each time it changes hands. Keep those pieces in one place and each handoff starts where the last one stopped.
ALT A chain of tools, ticket, Slack, Jira, Sentry and doc, with people at each step repeating 'So the customer...', followed by a single shared place where the handoff continues.
6
Good prompts at work often live in one person's chat history. If a call-prep prompt works, write it down as team instructions: which sources to read, cite the date of each fact, flag conflicts instead of picking one. Then anyone can run it and the reviewer knows what to check.
ALT A handwritten index card titled CALL-PREP team instructions: read CRM notes, last 3 tickets and the contract; cite the date of each fact; if two sources disagree, flag it; one page with open questions last. Three people beside it, captioned: anyone can run it, and reviewers know what to check.
16
Shantanu Vishwanadha retweeted
Day 2 of #WebSummit2026 canceled because of Storm Charles, so 70k tech people got an impromptu rain day in Lisbon.
Spent the morning in a hotel lobby planning dinner for our group across different hotels, all inside Hops.
Thread pulled everyone’s locations into a shared map, set up a ranked-choice voting table, and tallied the picks in under 15 minutes.
Real collaborative AI is built for messy group workflows like this.
@hops_ai @WebSummit
A DevRel use for a shared AI workspace: have it read the week's GitHub issues and community threads, then list the questions the docs don't answer, each linked to its thread. The docs team edits that list in the same place and can open any original thread.
ALT This week's GitHub issues and forum and Discord threads funnel through AI into a list titled Questions the docs don't answer, each item linked to its issue or thread, with docs and devrel cursors editing it.
24
I use Hops to prepare demos of Hops. At this point it knows the pitch better than I do.
Support tickets, video scripts, client demo prep: it all lives there now, and I've stopped losing my train of thought every time I switch tabs.
As of today, it's open to everyone.
Want a walkthrough? My DMs are open. I'll hop on a call with you, and I'll bring Hops along. It's had plenty of practice.
🤖 Made with AI
1
46
Shantanu Vishwanadha retweeted
AI giants are spending billions on personal assistants.
I think that's the wrong bet for companies. The work that matters happens in teams.
Today we launch Hops, where teams and AI agents work together in shared chats, docs, tables, slides and apps.
For AI features, a support request is rarely only a bug report. It can expose an unclear default, a missing constraint, or a workflow that asks users to guess. Treating those signals as product input is part of adoption work.
34
The most useful AI documentation is often a decision boundary.
“Use this when…” and “Don’t use this when…” save more time than a larger feature list. The second line gives people permission to stop forcing the tool into a problem it does not fit.
33
AI support should optimize for the next action, not the longest explanation.
A good answer tells a developer what to try, what evidence to collect if it fails, and where to go next. That is more useful than a polished summary of every possible cause.
32
For developer tools, activation is not a first successful prompt. It is the moment a developer knows when to trust the tool, when to verify it, and how to recover when it is wrong.
37
The dangerous AI demo is the one where output looks finished before anyone asks whether it fits the repository, the policy, or the user's intent.
Fluency shortens review only when provenance and reversibility are part of the workflow.
1
50
AI answers in support should be judged by the decision they enable, not their tone.
A concise case summary with product area and evidence is often more valuable than a perfect reply. It lets a human decide whether to answer, escalate, or fix the docs.
38
Developer documentation is part of an AI system's control surface. When the canonical answer is stale, fragmented, or hidden behind access, retrieval reproduces that confusion at speed. Fix the source before tuning the prompt.
2
39
A support assistant needs a visible "I can't tell" path. If it cannot point to the source, state uncertainty, or hand off cleanly, it turns missing knowledge into confident prose. That is worse than a slow search.
49
For practical AI adoption, time-to-first-draft is a weak metric. The useful metric is time-to-verification: how long it takes someone to decide the output is safe to send, ship, or act on. If verification stays slow, the workflow has not improved much.
2
2
62
I’m skeptical of AI features that hide uncertainty to sound decisive. A concise answer with visible sources or a clear “I don’t know” gives users something they can check. Confidence without a path to verify it creates more support work later.
45
For developer tools, prompts are part of the interface. Changing a prompt can alter behavior as much as changing an endpoint. Version them, test representative tasks, and record what changed. “We improved the prompt” is not a release note.
1
49