Executive summary
Software is leaving the screen. The next decade of productivity gains will not come from a better dashboard but from no dashboard at all - work that happens inside the channels people already live in, surfaced as a message rather than a destination to visit. We call this category invisible software, and we believe it serves two markets that look opposite and are in fact the same bet.
At one end sits operational necessity. The deskless workforce - field technicians, drivers, nurses, line leads - cannot carry a screen and was never well served by one. For them, a text-first interface is not a preference; it is the only interface that survives gloves, motion, and a shift floor. Invisible software here is adoption by physics.
At the other end sits luxury efficiency. The high-value desk worker - the founder, the partner, the operator - can run twenty tools and increasingly refuses to. Their scarcest asset is attention, and every open tab is a standing tax on it. For them, invisible software is the luxury of a closed laptop: the inbox and the calendar run themselves through a single conversation while the screen stays dark.
Destination software asks you to come to it. Invisible software comes to you - and the winning layer is the one that orchestrates the others, not the one with the prettiest screen.
The structural move is from destination software to invisible software: orchestration layers that live in SMS, Slack, and email, read and write across the underlying systems of record, and return a single clear answer. The interface collapses to a conversation. The rest of this memo sets out the problem this solves, the thesis behind it, the two market segments we would underwrite, and the framework we would use to source and evaluate companies in the category.
The problem: tab fatigue and GUI failure
The graphical interface won, then quietly began to fail at scale. The average knowledge worker now keeps a standing herd of tabs open all day, each one a small, permanent claim on attention. The cost is no longer the software; it is the friction of moving between it.
The context-switching tax
Every jump between tools forces a reload - not of the page, but of the person. Re-finding context, re-orienting, re-deciding where you were. This tax is paid dozens of times an hour and never appears on any invoice. It is the single largest hidden cost of the modern stack, and adding tools to feel more in control only raises it.
Dashboard blindness
The reflexive answer to complexity has been another dashboard: one more place to look, one more set of charts to interpret. But a dashboard that must be visited is a dashboard that gets ignored. Signal that requires a deliberate trip to find is signal that arrives late, and most of it is never acted on. The screen meant to create clarity becomes one more surface to monitor.
The desktop anchor
Most software still assumes a worker tethered to a desk and a large screen. That assumption excludes the majority of the global workforce outright and constrains the rest. The desktop anchor is why deskless teams run on paper and phone calls, and why executives feel chained to a laptop to do work that is, at its core, a series of small decisions. Remove the anchor and the decisions can happen anywhere - including over text.
The thesis: invisible software and zero-UI
We believe the dominant interaction model for a large class of work is moving from screens to conversation. Invisible software, sometimes called zero-UI, does not replace the systems of record beneath it; it orchestrates them and presents one human-legible surface on top - the message.
Communication as interface
The interface becomes the channel the user already lives in: a text thread, a Slack DM, an email. There is nothing new to learn and nothing new to open. The instruction is plain language - “move my Thursday”, “approve the vendor” - and the response is a single confirmation. The conversation is the application.
Proactive push over reactive pull
Dashboards are reactive: they wait to be pulled. Invisible software is proactive: it pushes the one thing that needs a decision, when it needs it, and handles the rest silently. The user is interrupted only by what is worth an interruption, and never has to go looking. This inversion - push the decision, absorb the busywork - is the source of the felt magic.
The orchestration layer
The defensible position is not the channel and not the model; it is the orchestration layer between them and the systems of record. It reads from the inbox, the calendar, the CRM, and the operational backend, decides what matters, takes action across them, and reconciles the result into one reply. Owning that layer means owning the relationship with the user and the write access to everything beneath - the position from which a durable company is built.
The value accrues to whoever sits between the conversation and the systems of record. That seat is the company.
Market segments
Two segments adopt invisible software for opposite reasons. One is pulled in by necessity, the other by the pursuit of efficiency. We would underwrite both, and watch for the founders who can serve either without building two companies.
The operational frontline
Frontline logistics, field operations, deskless service work
The largest, least-served labor market in the world runs without a screen. Invisible software reaches it through the device every worker already carries and the interface every worker already knows - the text message. Adoption is driven by physics, not preference, and switching costs compound as the layer becomes the system of record for the field.
- Adoption gated by hardware reality, not training
- Enormous, software-underpenetrated TAM
- Becomes the write layer for field activity
The luxury executive
High-value desk workers, founders, partners, operators
A smaller, high-willingness-to-pay market that treats the closed laptop as a status of control. Invisible software offloads the inbox, the calendar, and the follow-ups to a single thread, defended on the user's terms. Here the product is sold on reclaimed attention, and the moat is how precisely it learns one person's judgment.
- Premium pricing against the value of attention
- Sticky once it learns individual judgment
- Word-of-mouth distribution among peers
The segments share an engine. The same orchestration layer that logs a completed job for a technician can triage a board reply for a partner. A company that builds the layer once and points it at the segment with the cleaner wedge - then expands - is the shape of the outcome we are looking for.
Sourcing & evaluation framework
We would judge companies in this category against four technical moats. Chat is easy to demo and hard to defend; the durable value sits in the write layer, the integrations, the distribution, and the depth of the orchestration. A serious contender is strong on at least three.
- i
System-of-record write layer
The strongest position is the ability to write back, not just read. A company that becomes the layer through which actions are committed - jobs closed, calendars changed, approvals sent - owns the workflow and is structurally hard to displace. We test for genuine write access and the trust required to hold it.
- ii
High-integrity integrations
Invisible software is only as good as its connections to the systems beneath it. We look for integrations that are deep, reliable, and resistant to breakage - not brittle scrapers, but durable, permissioned access that survives the partner's roadmap and earns the right to act on the user's behalf.
- iii
Distribution velocity
Living inside SMS, Slack, and email is a distribution advantage when it compounds. We favor models with built-in virality or organic spread - a frontline worker who pulls in their crew, an executive whose peers ask what they are using - so that acquisition cost falls as the network grows.
- iv
AI orchestration depth
The reasoning layer is the differentiator that is hardest to copy. We assess how well the system plans across tools, recovers from ambiguity, and learns the individual user's judgment over time. Depth here is what turns a chat wrapper into an operator the user trusts with irreversible actions.
Investment posture
We would lead or co-lead at the seed and Series A stage in teams building the orchestration layer for one of the two segments above, with a credible path to owning the write layer and a wedge that spreads on its own. The category is early, the language is still settling, and the winners are unlikely to look like software at all.