Pinned post
Eric Seufert
Verified account This account is verified. Learn more@eric_seufert
Independent Analyst, MDM: mdm.co | Author, Freemium Economics (Elsevier) | Theseus: theseus-growth.com. Per commercium virtus.
2.1K Following 50K Followers 23K Posts
Latest posts 5
Apple should not buy Instinct.
Amazon just announced that it will allow customers to communicate with its own agent, Alexa for Shopping, via SMS. Apple already owns that interface.
Agents are commodities. Perplexity’s Computer, OpenClaw, Instinct, Muse. ByteDance will offer one, as will Google and Microsoft.
An agent can be built by a motivated team in a matter of weeks; no small development feat, but not an insurmountable hurdle by any means.
Distribution and privileged access to data / platform functionality is what differentiates agents. Apple has that, and buying Instinct provides it with no obvious strategic advantage that it couldn't build itself. Siri already has agentic functionality and can connect to iOS apps with the native App Intents API.
Everything one could argue about the logic of Apple buying Instinct applied 12 months ago with respect to Perplexity. Commodity agentic functionality is not inherently valuable; it’s unlocked with platform reach. M&A is not a prudent growth strategy in this space.
Amazon just announced that it will allow customers to communicate with its own agent, Alexa for Shopping, via SMS. Apple already owns that interface.
Agents are commodities. Perplexity’s Computer, OpenClaw, Instinct, Muse. ByteDance will offer one, as will Google and Microsoft.
An agent can be built by a motivated team in a matter of weeks; no small development feat, but not an insurmountable hurdle by any means.
Distribution and privileged access to data / platform functionality is what differentiates agents. Apple has that, and buying Instinct provides it with no obvious strategic advantage that it couldn't build itself. Siri already has agentic functionality and can connect to iOS apps with the native App Intents API.
Everything one could argue about the logic of Apple buying Instinct applied 12 months ago with respect to Perplexity. Commodity agentic functionality is not inherently valuable; it’s unlocked with platform reach. M&A is not a prudent growth strategy in this space.
Apple buying Instinct for $30 billion would be 0.6% of their market cap, basically a rounding error.View quoted post on X
They'd save 18-24 months of work plus a heap of execution risk.
Heck, if Apple really wants to roll their own, just buy Instinct anyway, as a hedge. Throw your weight around!
When I've been brought into AI transformation projects, they often take shape in three stages:
1) "Throw an LLM at it." Integrate an LLM or some sort of generative tool at every workflow. This is usually driven by fear; management thinks the company is falling behind on AI adoption, so every team is given a mandate to adopt some new AI-assisted tool, or to build one, and this can take the form of simply giving many members of the team large-limit Claude / Codex plans, to procuring enterprise contracts with those companies, to building proprietary tools that work through VertexAI, Bedrock, or AI Foundry.
2) "Efficiency benchmarks." The first stage becomes very expensive, so teams identify workflows that don't benefit from these models and revert to their prior operating models. Ironically, this is often interpreted as cost savings.
3) "Token diet." Teams start considering whether they need frontier models for workflows and tasks that are made more efficient by integrating LLMs or other generative models. What if we can use 3.8 Flash instead of 3.1 Pro? What if we can use 3.5 Flash instead of 3.8 Flash? This requires custom evals and is time-consuming, but it often delivers substantial cost savings with little to no trade-off in output quality.
If I had to guess, 20% of the companies in the consumer app space have even started this sequence. Maybe ~10% of those have progressed past stage one.
1) "Throw an LLM at it." Integrate an LLM or some sort of generative tool at every workflow. This is usually driven by fear; management thinks the company is falling behind on AI adoption, so every team is given a mandate to adopt some new AI-assisted tool, or to build one, and this can take the form of simply giving many members of the team large-limit Claude / Codex plans, to procuring enterprise contracts with those companies, to building proprietary tools that work through VertexAI, Bedrock, or AI Foundry.
2) "Efficiency benchmarks." The first stage becomes very expensive, so teams identify workflows that don't benefit from these models and revert to their prior operating models. Ironically, this is often interpreted as cost savings.
3) "Token diet." Teams start considering whether they need frontier models for workflows and tasks that are made more efficient by integrating LLMs or other generative models. What if we can use 3.8 Flash instead of 3.1 Pro? What if we can use 3.5 Flash instead of 3.8 Flash? This requires custom evals and is time-consuming, but it often delivers substantial cost savings with little to no trade-off in output quality.
If I had to guess, 20% of the companies in the consumer app space have even started this sequence. Maybe ~10% of those have progressed past stage one.
I don’t think I’ve ever encountered writing like this in The Economist before.
How big is Google’s AdMob? mdm.co/how-big-is-goo…
mdm.co How big is Google's AdMob? | MDM by Eric Seufert How big is Google's AdMob?