Signals & Systems is a series sharing straightforward insights on technology, leadership, and day-to-day execution, directly from the operator’s perspective. Written by OAG COO Filip Filipov, each edition decodes the signals pointing to the future of tech and the systems we're using to get there.

In line with all Europeans, I took a bit of time off in August - but now we are back to it, rotating between system thinking and signals from the world of AI and travel. Today, we cover a combination of both:

1 | Signals - What Moved

Latency is going down across all foundational LLMs, allowing multi-agent orchestration without lag. What it means is that you can get answers (inference) faster than before, which speeds up the entire stack. Speaking of stack, it seems to be collapsing in a few interfaces.

• OpenAI's latest ChatGTP release is brilliant, as it now decides on its own whether to go System 1 (fast) or System 2 (slow) thinking. While it seems to be dismissed as 'not enough innovation', I think it is a strong way to triage how to answer the question. Beyond the point, though - OpenAI is going vertical - the hidden messages that the team is doubling down on Health assistance is a signal that in the future, you'd go after the verticals: shopping, travel, fitness coaching, writing, etc.
• Perplexity is not only pushing for its own browser Comet, but also making a bid for Chrome, in case it is for sale (to the tune of $33B)
• Google is already releasing AI and chat-like capabilities across its product verticals - what started as search is now shifting to travel (Flights is the first one, with releases in the US and India)

Why it matters: For the first time, distribution and product are converging. If previously search and social were distribution to take you to the product experience, and even today ChatGPT drives strong traffic to sites, ultimately, why should one go to a site to read the same information. Potentially massive consequences for product people and the discipline in general. 

2 | System: The Scaling Coefficient

Scaling a VC-backed startup is different from scaling an established PE company. In both cases, we can agree that roughly (and this changes and is not scientific), once you get to venture scale (roughly $100m ARR), the journey to $1Bn in annual revenue is what constitutes scaling. After that, you are actually a big player in your field.

So how do you measure your success in this area? For VC-backed startups, growth is what matter (less so unit economics), while for PE-backed companies, the ideal is to focus on strong, sustainable, and profitable growth.

After quite a bit of thinking about the topic, I distilled it to a composite measure - the Scaling Coefficient (I am sure there can be a better name).

Scaling Coefficient = Revenue per Employee × Rule of 40 (YoY Growth % + EBIT Margin %)

It fuses three truths into one number: growth × profitability × output per seat.

Why now?

• Agents are shifting work from “more people” to “more capability per person”
• Latency collapse means multi-agent workflows actually stick in daily ops
• Bundling (email, docs, calendar, browser, other productivity tools) concentrates value in fewer surfaces — SC tells you if you’re capturing it

How we as leaders can use it

• Compass, not trophy. Track SC level and change (q/q, y/y); discuss why it moved (price/mix, adoption, infra cost)
• Resource filter. Fund bets that raise RPE or Rule-of-40; time-box those that don’t.
• Per-seat playbook. Make “AI-first professional” the default; instrument how each team’s RPE evolves post-agent rollout

Guardrails

While no measure is perfect, there needs to be a balance against customer health (NRR, churn, etc - clearly, you can game the number by booking short-term revenue) and logic - dropping the Scaling Coefficient because you are investing in talent means you'll see a lag in the overall number, while your EBIT drops for a period of time.

Yet, by keeping this number in mind, you can see efficiency, growth, and profitability and you can easily keep constant the overall goal - measuring output of the organisation.

In subsequent posts, I will test the hypothesis against some public companies to see if it truly sticks, but as a temperature check of your ability to scale sustainably and profitably, it does stand the sniff test for now. (would love to hear your thoughts to make this argument better).

Why it matters: Scaling Coefficient can be the board-level signal for the agent era. I’ll publish a deeper cut next: methodology, benchmarks, and how to build it tile into your weekly ops review.

3 | Operator’s Radar

Here's a few things that I've been reading/listening/caught my eye in the last couple of weeks.

• A brilliant podcast by Lenny Rachitsky talking to Eoghan McCabe in how founder mode can be brought back even in established companies: hard, albeit unpopular decisions, strong conviction, moving at speed. A dream. https://youtu.be/0_opWSfmN8M?si=1SQIG_lUEzDQfNqj. Also, I highly recommend the podcast for all leaders, not just in product, so subscribe!
• Chris Williamson's brilliant Modern Wisdom podcast has so much in it that catching up with old, timeless episodes is a delight. This one is on confident speaking with Matt Abrahams - take away as a useful framework: what, so what, now what - https://chriswillx.com/podcast/
• Another playbook by Madhavan Ramanujam and a must-read for all C-level execs on pricing for scale - Scaling Innovation https://www.amazon.com/Scaling-Innovation-Companies-Architect-Profitable/dp/1119633060 

4 | Coming Up

Systems: Guarding the Swarm - a pragmatic playbook for agent risk & governance. I skipped it for this edition, as I thought I needed some more thinking on the topic. 

What an exciting time to enter the last part of 2025. 

Onwards!