The large language model (LLM) market has become one of the most crowded and capital-intensive arenas in technology. Estimates peg the sector at $5.6 billion in 2024, on track to reach $7.36 billion this year and potentially as much as $35 billion by 2030, growing at a compound annual rate of over 35%. Some analysts push those forecasts even higher, projecting a $130 billion market by 2034. Against that backdrop of billion-dollar burn rates, GPU shortages, and the race toward “everything machines,” Writer.com is pursuing a different strategy.
As Dan Bikel, Writer’s new head of AI, explained to me at the AI4 Conference, “Writer doesn’t have to be all things to all people. It only needs to deliver amazing systems and models powering the enterprise”.
Systems, Not Just Models
Bikel comes to Writer with deep research experience and a scientist’s instinct for skepticism. His remit is to guide the company’s Palmyra model series and expand initiatives like self-evolving models and action agents. The former lets models “reflect on their own output in such a way that [they] can then generate training data for [themselves] that will help [them] improve in the future”. The latter takes enterprise automation a step beyond RPA, letting companies specify problems instead of workflows and having Writer’s AI agents orchestrate the solution.
What’s notable is how Bikel frames the mission. “I’ve already told my team—it’s about systems, not models. We are all responsible for the systems that we ship, and that means taking responsibility for everything around them”. That contrasts with the LLM giants, where scale—encompassing more data, larger clusters, and bigger models—remains the prevailing strategy.
From Meta to Writer
Bikel’s credibility isn’t theoretical. Before joining Writer in July, he spent years at Meta, working on some of the most challenging problems in natural language processing. At Meta, he was part of a research culture that tackled challenges such as safety, multilingual translation, and social conversational AI. That background gave him a front-row seat to the sheer scale—and limitations—of frontier models when deployed to billions of users. At Meta, the emphasis was often on building systems that could “delight and entertain and emotionally support”. But as Bikel noted, those objectives are not what enterprise customers demand. That distinction is key: his move to Writer reflects a deliberate shift from building for consumers at internet scale to building for businesses with specific, high-stakes workflows.
The question, of course, is how Writer survives in a world where OpenAI, Anthropic, Google, and xAI are spending billions on GPUs and researchers. Bikel’s answer is deceptively simple: focus. Writer doesn’t chase AGI. It doesn’t need to cover “vast swaths of the human condition.” Instead, it targets enterprise clients—companies that want models tuned for productivity, compliance, and measurable ROI, not chatbots built for emotional support.
That focus shapes everything from training data (“we don’t want everything under the sun because we’re not interested in all emails”) to benchmarks. Writer’s Palmyra series, he notes, has tested “almost exactly at the average of the top five scoring models” on HELM, one of the industry’s respected evaluations. In other words, Writer is competitive on quality while avoiding the ruinous costs of frontier scaling.
What truly sets Writer apart is philosophy. “One of the phrases that Writer likes to use is that they have a human-centric vision for AI … it’s enabling people to do things that they should be able to do to make their business thrive”. That ethos resonates at a moment when many enterprises are wary of handing mission-critical tasks to opaque black boxes. Writer’s model is built around accountability, transparency, and practical augmentation rather than the pursuit of artificial general intelligence.
That strategy also explains how Writer can thrive in an enterprise ecosystem dominated by giants. Salesforce, for example, is already a Writer customer. For a company that already has its own AI stack, choosing to plug in Writer demonstrates that Writer’s tools fill fundamental gaps. As Bikel put it, “Where they see gaps, they can see [Writer’s] stack … and maybe they’ll want more”. For younger firms without deep AI capabilities, Writer’s end-to-end platform—complete with agent building and hosting—becomes even more appealing.
This positioning allows Writer to avoid the arms race for ever-larger models and instead compete where it matters: delivering value to enterprises that need AI aligned to their workflows and governance requirements.
Punching Above Its Weight
Bikel is quick to acknowledge the odds. Writer is not the best-funded player in the game. It doesn’t have tens of thousands of GPUs on order or a global research staff numbering in the thousands. But that’s exactly why the company intrigues him. “Writer is punching above its weight. It’s not trying to replicate ChatGPT—it’s succeeding by focusing on enterprise needs and building systems that actually work”.
With the LLM market barreling toward tens of billions in value, Writer’s approach is refreshingly contrarian: smaller, more curated, relentlessly enterprise-focused. It may never be a household name—but in a field where commoditization looms, that might be the most innovative strategy of all.
That’s also why Bikel’s career move matters. At Meta, he saw firsthand how consumer-facing AI risks becoming a commodity—ubiquitous, expensive to maintain, and ultimately undifferentiated. By joining Writer, he’s betting that the real opportunity lies not in building ever-larger general models but in crafting systems that solve concrete business problems. In a market where the giants may soon undercut each other into sameness, Writer’s focus on human-centric, enterprise-ready AI looks less like a niche and more like a hedge against commoditization itself.
