Commercial workflows move online

The interesting thing about June 2020 is not that commercial work moved online. It is that it moved online with enough pressure behind it to expose what was actually missing.

Most businesses had some digital infrastructure before the spring. Email, shared drives, the occasional project board. What they did not have, what many are only now discovering they lack, is an actual operating system for commercial execution. The tools were there. The system was not.

Everything that was holding together stopped holding

There is a particular kind of organisational knowledge that lives in proximity. The salesperson who knows which deals are warm because they sat next to the account manager last Tuesday. The creative director who knows the brief has changed because she happened to overhear a client call. The operations lead who knows to check the spreadsheet in the shared drive, but only after someone has manually updated it, which usually happens on Thursdays.

This knowledge does not survive a move to fully distributed working. It cannot be forwarded in an email or reconstructed from a task board. It depends on the social architecture of a shared space, and when that architecture was removed in March 2020, it took a great deal of operational coherence with it.

What replaced it, in the organisations that managed the transition without collapsing, was a mixture of increased communication volume and hastily assembled tool stacks. More calls. More messages. More status updates. The underlying workflow gaps remained; teams were simply compensating for them more actively. That is an exhausting way to operate, and it is not stable. The productivity cost of that compensation was not always visible in the short term, but it was real.

By June 2020, the emergency phase is mostly over. The companies that were going to fold under the pressure have largely done so. The survivors have made choices about which tools to rely on, which processes to formalise, and which habits to build. The dust has not entirely settled, but the shape of the new baseline is becoming visible. And what that baseline reveals is an enormous amount of unfinished work in commercial software.

The difference between moving online and operating online

There is a meaningful distinction between moving online and operating online, and it matters for understanding where the real product gap is.

Moving online means that the activities which used to happen in a physical space now happen through a screen. Sales calls through video conferencing. Creative reviews through shared documents. Client updates through email rather than in-person meetings. This is what most organisations managed to accomplish between March and June 2020, and it is the version of "going digital" that gets most of the attention.

Operating online means something harder. It means that the logic of commercial execution, how a business finds clients, develops opportunities, scopes and prices work, coordinates delivery, and learns from what it has completed, is encoded in software rather than in the accumulated habits of people who have worked together for a long time. It means the workflow has a memory. It means handoffs are built into the system rather than managed by whoever happens to be paying attention. It means the operating surface is the business, not a set of disconnected tools bolted together with communication overhead.

Most businesses have achieved the first thing. Very few have achieved the second. The gap between them is where the product opportunity lives, and the conditions of 2020 have made that gap significantly more visible than it was before.

The arc that no single tool covers

Think about the actual shape of commercial work for a services business, a studio, a software company, a professional practice: any organisation where the output is a delivered engagement rather than a physical product.

It starts before there is a client. There is a lead, a conversation, a relationship being developed. Someone has to track that. Someone has to know where it is in the sequence, what the next action is, what the context of the relationship is. That is one surface.

Then there is scoping: understanding what the work actually is, what it will cost, whether it is worth doing, how to price it in a way that reflects its real value. That is another surface, and it connects back to the first.

Then there is delivery. Work being organised, assigned, tracked, adjusted when things change. Clients being managed. Feedback being incorporated. Timelines being maintained or renegotiated with intelligence rather than panic. That is a third surface.

And then there is the close. Invoicing. Handover. The reflection on what happened. The information that, if captured properly, would make the next engagement easier to scope, price, and deliver. If it is not captured, it disappears, and the business starts from scratch every time.

No CRM covers that arc. No project management tool covers it. No combination of generic tools assembled into a dashboard covers it. What covers it is a purpose-built operating system for commercial execution: one that holds the complete workflow and provides a coherent surface for managing it.

That is the space Orbit is being built to occupy. Not as a feature added to something else, not as a CRM with extra fields, but as a genuine operating surface for the full lead-to-launched-product workflow. The conditions of 2020 have not changed the thesis. They have made it more legible.

TUXX in live conditions

TUXX sits in a particular position relative to this moment. As the services and custom systems arm of the portfolio, it is operating in the actual conditions commercial teams are navigating, not in hypothetical versions of those conditions, but in live engagements with real clients, real constraints, and real commercial work to execute.

That proximity is instructive in ways that desk-level product thinking rarely is. When you are helping a client migrate a workflow that has lived in email threads and muscle memory for years, you encounter the specific points at which that migration breaks down. The handoffs that do not translate into software without significant redesign. The status information that everyone assumed was shared but was never actually recorded anywhere. The approval steps that existed because of one difficult client experience two years ago, and which nobody has thought to revisit since.

These failure points are not random. Across different clients, different industries, different scales of organisation, the same structural gaps appear. The same workarounds get rebuilt. The same compensating behaviours, the extra call, the manual update, the informal check-in that should not need to be informal, emerge to patch the same underlying absence.

That repetition is the signal. It is evidence about where the operating system is missing, and therefore about where product development needs to go. It does not translate directly into a feature list, client patterns rarely do, but it informs the conviction about what a properly built commercial operating surface needs to handle.

GPT-3 and the question of what the capability is for

June 2020 also saw the release of GPT-3, described in its accompanying paper as a large language model trained at a scale that produced qualitative improvements in language task performance without task-specific fine-tuning. The model demonstrated that increasing scale, in parameters and in training data, shifted the baseline of what a language system could do across a wide range of tasks.

The honest position, in June 2020, is that the precise commercial implications of this are not yet clear. The paper itself is careful about what it claims. What is clear is that the direction is significant, and that language processing at this level of capability will eventually matter for the kind of work being described in this piece.

Commercial workflows are saturated with language tasks. Drafting proposals. Summarising client briefs. Classifying incoming requests. Routing work to the right people. Writing follow-up communications. Generating status updates from underlying data. These activities consume time and attention throughout the commercial arc, and they are precisely the kind of activities that capable language processing could begin to handle differently.

The question is not whether that becomes relevant. The question is what kind of surrounding architecture makes it useful. A model that can draft a paragraph is not the same as a system that knows what paragraph needs drafting, in what context, for which client, at what stage of which engagement. The capability requires context. Context requires memory. Memory requires a system that has been built to hold and organise the commercial workflow in the first place.

This is why the operating system question comes before the AI assistance question. The intelligence layer, what Orion is being built to become, needs a coherent commercial surface to be intelligent about. Without that surface, capability without context is a party trick. With it, it becomes a genuine change in what commercial execution can look like.

What the moment is actually asking for

The dominant narrative around commercial disruption in 2020 is one of acceleration: everything that was going to happen eventually is happening now, faster than expected. There is truth in that framing, but it can obscure something more specific.

What is actually being asked for, in June 2020, is not more tools. Businesses have more tools than they can effectively use. What is being asked for is coherence: an operating logic that holds the commercial workflow together rather than fragmenting it across a set of disconnected surfaces that do not share memory, context, or state.

The businesses that will come out of this period in a genuinely stronger position will not be the ones that adopted the most tools. They will be the ones that found or built systems that made their commercial execution more coherent. That is a different kind of problem, and it requires a different kind of solution.

The work is in building that solution. Not for a disrupted moment. For commercial execution as it needs to work when it lives entirely in software.

That is what this period is clarifying. And that clarity is worth building from.