The setting
makes productivity software for email-heavy workers and teams. is an AI assistant that takes on tasks across the browser, desktop and Superhuman apps, working from goals written in plain language.
Recent client commitments are competing with a larger volume of older conversation history.
Briefing quality depends on finding the latest decision without making retrieval slower or more expensive.
Who you’d be doing this for
“The client asked why we were discussing an option they’d already declined yesterday.”
Dawit Ndirangu · Agency Account Director · a character in this brief
Prepares client briefings from long email histories before status calls.
What is at stake
Agency account directors are opening outdated threads before client calls. You must weigh recency against relevance without slowing every briefing request.
Why it isn’t already fixed
Every obvious fix costs something else. That’s the part you’d have to decide.
- recency signals vs. decision relevance
- briefing trust vs. retrieval speed
- shared ranking consistency vs. surface-specific delivery
- model quality vs. serving cost
Why Superhuman
Superhuman Docs and Grammarly place Go in the moment client-facing workers need current context before an external conversation.
Written with these in mind
Not your kind of problem? 36 more at Superhuman, or browse every organization.
Where the organization context comes fromThe postings this brief was written from, and the reporting read before it.
The organization context comes from public information. The situation and characters are written for the brief.
Job postings it was written from
Public descriptions of the work at Superhuman. They are listed as sources, not as openings. The lines under a posting are its own words, where it meets this brief.
- Machine Learning Engineer (opens in a new tab)
- You'll own how the agent plans the work, picks the right agent or tool for each step, and sees long tasks through to the end.
- Own Critical ML Systems: Develop and deploy sophisticated ML models for core product experiences, including search ranking and proactive suggestions that anticipate user needs.
- Senior Design Technologist, AI Tooling (opens in a new tab)
- You will help ensure our AI-native platform (spanning email, calendars, docs, and agents) is supported by scalable systems that maintain our high-quality standards.
- AI Engineer (opens in a new tab)
- Build high-quality evaluation datasets: Design, source, and curate golden sets, labeled data, and synthetic data for agent behaviors, search ranking, and proactive suggestions.
- As Superhuman Docs’ collaborative workspaces evolve so users can solve arbitrarily complex tasks with agents, rigorous measurement will be what lets us ship with confidence.
- Software Engineer, Full-Stack (opens in a new tab)
- Build a Groundbreaking AI Agent Platform: Spearhead the development of a new AI platform inside the Superhuman Docs product of the Superhuman suite where multiple AI agents collaborate to solve complex user problems.
This is the setup. The work is inside.
Running it puts you in the room: the full situation and its constraints, stakeholders who push back in their own words, and the decisions that are yours to make. What you produce becomes a Day One Plan — work you can show someone instead of describing.
Where similar work appears
Other live briefs connected to this one, and what connects them.
Similar problems at other organizations
Like this one, each of these turns on deciding on evidence that's still thin.
Blocked-work queries are surfacing old project updates instead of actionable tasks.
You have to weigh ranking precision against retrieval latency and context cost.
Only 54% of affected searches produce a top-three result click, and operations managers are returning to source apps for current files.
You must weigh a targeted relevance gain against keeping discovery strong for every other search.
Eighteen ML engineers reported manual source selection and widely uneven answer grounding.
You must weigh a useful first ranking model against evidence that is incomplete across mixed enterprise assets.
The same setting, from another role
Each is set around Superhuman’s Superhuman Go, written for a different role. They’re separate briefs, not the same scenario.
Go workflows complete in the Chrome extension and then stop after a user continues in the desktop app.
You must decide whether the system or the engineers decide which of streaming state, context gathering or the surface handoff gets treated as the cause when they disagree.
Enterprise users can approve prepared work without one complete view of its connected reads and writes.
You have to weigh a fast approval flow against the detail administrators and client-facing users need to trust it.
A replay showed Go preparing an update from context after an administrator removed access.
You have to weigh a fail-closed control against valid workflows that may be blocked during an enterprise rollout.
Other problems with a similar purpose
In each of these, as in this one, the aim is to make a machine-driven decision dependable and fair at the exact real-world moment a customer is counting on it.
Agents are retrieving older interaction summaries after a customer updates a preference.
You must weigh a ranking correction against retrieval latency and safe rollout coverage.
Fourteen percent of Qwen requests now exceed three seconds after the container build.
You must weigh faster batching against the context available to each answer.
Clinicians are rejecting exercise-progression recommendations because the agent retrieves outdated protocol content.
You must weigh a fast retrieval correction against regression risk across shared deployments.
