PromptQL is a spin-off from Hasura pitched as an agentic workspace that replaces the ambient chatter of Slack with a system where humans and AI agents share the same context. The product connects to sources like chat, code, and meeting recordings, then builds a persistent map of what teams have discussed and decided so agents can act on it. In the launch demo, a project management agent traces a blocked engineer's work across pull requests, threads, and meetings, finds the missing copy and design assets, pings the right teammate, and orchestrates other tools to unblock the task without the founder relaying messages by hand.
The launch matters now because agent tooling is shifting from single prompts to long-running work that needs organizational memory, and PromptQL is betting that the memory layer belongs in the workspace itself rather than inside individual bots. The pitch to founders and operators is a stack that captures process context as it happens, reuses "skills" learned from past work (like a Salesforce integration built by a senior engineer), and hands finished drafts back for human review before shipping. The launch tweet claims customers including Instacart, McDonald's, and Cisco, and the team is giving away a bundle of prebuilt skills and agents to people who reply to the announcement.
PromptQL is led by Tanmai Gopal, co-founder and CEO, who previously led the creation of the Hasura GraphQL Engine , alongside co-founder Rajoshi Ghosh . The company grew out of Hasura's work on data access for AI systems and is based in San Francisco, with the founding team drawing on the same background that took Hasura's open-source engine into wide enterprise use.
Comments (7)
Sign in to join the discussion.
P
Priya KothariJul 8, 2026
"We raised $136M to kill Slack" is a top-tier opening line, but the follow-up video buries the actual demo at 90 seconds in. Cut the intro, lead with the shared context view, watch the retention curve fix itself.
L
Lukas B.Jul 8, 2026
Killing Slack is like killing email in 2014. The graveyard is full of tombstones with cute logos on them.
O
obiomaJul 8, 2026
ok wait, the tagline is actually the thesis and the product name is the boring part. rename the company Second Job and I'll pre-order a hoodie.
M
Meiko TanabeJul 8, 2026
Does the agent have an API for reading channel context, or is it purely UI-driven? Asking because I already have three Slack bots that will absolutely fistfight yours over who owns the standup thread.
F
Fern LacroixJul 8, 2026
Rewriting the tweet for free: "Slack remembers nothing. PromptQL remembers everything." You're welcome, invoice in the DMs.
R
Raoul M.Jul 8, 2026
We built a version of this at my last job in 2019, we called it "a shared Google Doc pinned to the channel." Genuinely curious what makes the agent version stickier than a wiki nobody updates.
Y
yesenia q.Jul 8, 2026
hot take: the number in the tweet is doing 80% of the work here. if this said $2M seed nobody would click, and yet the product might be the same product.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
PromptQL
Maintaining context is the second job nobody wants.
Maintaining context is the second job nobody wants.
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
PromptQL is a spin-off from Hasura pitched as an agentic workspace that replaces the ambient chatter of Slack with a system where humans and AI agents share the same context. The product connects to sources like chat, code, and meeting recordings, then builds a persistent map of what teams have discussed and decided so agents can act on it. In the launch demo, a project management agent traces a blocked engineer's work across pull requests, threads, and meetings, finds the missing copy and design assets, pings the right teammate, and orchestrates other tools to unblock the task without the founder relaying messages by hand. The launch matters now because agent tooling is shifting from single prompts to long-running work that needs organizational memory, and PromptQL is betting that the memory layer belongs in the workspace itself rather than inside individual bots. The pitch to founders and operators is a stack that captures process context as it happens, reuses "skills" learned from past work (like a Salesforce integration built by a senior engineer), and hands finished drafts back for human review before shipping. The launch tweet claims customers including Instacart, McDonald's, and Cisco, and the team is giving away a bundle of prebuilt skills and agents to people who reply to the announcement. PromptQL is led by Tanmai Gopal, co-founder and CEO, who previously led the creation of the Hasura GraphQL Engine , alongside co-founder Rajoshi Ghosh . The company grew out of Hasura's work on data access for AI systems and is based in San Francisco, with the founding team drawing on the same background that took Hasura's open-source engine into wide enterprise use.
Comments (7)
"We raised $136M to kill Slack" is a top-tier opening line, but the follow-up video buries the actual demo at 90 seconds in. Cut the intro, lead with the shared context view, watch the retention curve fix itself.
Killing Slack is like killing email in 2014. The graveyard is full of tombstones with cute logos on them.
ok wait, the tagline is actually the thesis and the product name is the boring part. rename the company Second Job and I'll pre-order a hoodie.
Does the agent have an API for reading channel context, or is it purely UI-driven? Asking because I already have three Slack bots that will absolutely fistfight yours over who owns the standup thread.
Rewriting the tweet for free: "Slack remembers nothing. PromptQL remembers everything." You're welcome, invoice in the DMs.
We built a version of this at my last job in 2019, we called it "a shared Google Doc pinned to the channel." Genuinely curious what makes the agent version stickier than a wiki nobody updates.
hot take: the number in the tweet is doing 80% of the work here. if this said $2M seed nobody would click, and yet the product might be the same product.