Cohere Compass 2 vs Perplexity Comet Browser

“The primitive here is a retrieval pipeline that can ingest mixed-format documents — PDFs with embedded charts, scanned tables, image-heavy slides — and return semantically relevant chunks without requiring a preprocessing ETL step per modality. That's a real problem: anyone who's tried to build RAG over a 10,000-document enterprise library knows the pain is 80% in the ingestion layer. The DX bet is that Cohere handles the multimodal parsing so you don't glue together a PDF parser, a table extractor, and a vision model yourself. The on-prem deployment option is actually the headline feature for the buyer, not the multimodal part — that's what gets it past legal review. My skip concern is documentation: the blog post is long on capability claims and short on API surface, schema design, and what 'image understanding' means at query time versus index time. Show me the query API, then we'll talk.”

“The primitive here is a Chromium wrapper with a Perplexity agent running over the DOM — form-filling and data extraction are just browser automation with an LLM deciding the selectors, which Playwright plus any capable model can do today without giving up your entire browsing session to a single vendor. The DX bet is that bundling the browser and the agent reduces integration friction, but that only pays off if you're a non-developer end user; any engineer is going to look at $50/month and immediately write a 200-line script. The moment of truth is asking it to log into a multi-factor authenticated enterprise portal and extract a report — that's where the walls appear. I'll ship when there's a headless API mode, a documented extension system, or evidence the agent handles real-world DOM chaos; right now it's a beautiful demo that hasn't published its error rate.”

“The direct competitors are Azure AI Search with multimodal indexing, AWS Kendra, and increasingly any RAG stack bolted onto GPT-4o's native PDF vision. Compass 2's real differentiator is not the multimodal capability — every major cloud provider is shipping that — it's the on-premises deployment for enterprises with data residency requirements, combined with a retrieval model trained specifically for enterprise document retrieval rather than general web content. The scenario where this breaks is at the 'chart understanding' claim: interpreting a bar chart semantically in a way that survives a specific quantitative query ('find all documents where Q3 revenue exceeded Q2') is a much harder problem than the blog post implies, and I've seen this class of tool hallucinate chart data confidently. What kills this in 12 months isn't a competitor — it's that the chart and table comprehension doesn't hold up under production query loads and the feature gets quietly deprioritized. I'm shipping it narrowly: for text-heavy PDFs with some visual elements in air-gapped environments, this is probably the best available option right now.”

“The direct competitor here is Arc Browser plus any AI extension, or just Chrome plus the Perplexity extension that already exists — and Perplexity already ships that extension. The specific scenario where this collapses is enterprise adoption: IT departments don't swap default browsers for waitlist products, and consumers don't either without a compelling reason beyond 'the sidebar is better.' The prediction: Google ships Gemini natively into Chrome at a depth Perplexity can't match within 18 months, and the browser angle becomes indefensible. For this to earn a ship, Comet needs a capability that is literally impossible to replicate in an extension — and form-filling and summarization are not that.”

“The buyer is a VP of IT or Chief Data Officer at a regulated enterprise — financial services, pharma, government — and the budget comes from the data infrastructure or compliance line, not a software tools budget. That's a real check-writer with a real problem: they have document libraries they legally cannot send to OpenAI's API, and they need search that works across formats. The on-prem deployment option is the actual moat here, not the multimodal capability — Cohere has been building that distribution channel for two years and it creates genuine switching costs once it's integrated into an enterprise's document management stack. The risk is that the pricing model is 'contact sales' all the way down, which means a long sales cycle and high CAC that has to be recovered on large contracts. What survives the model-gets-cheaper scenario is the enterprise integration layer and compliance certifications, not the retrieval model itself — Cohere needs to be pricing for that, not for compute.”

“The buyer here is unclear in a way that should worry everyone: consumers don't pay for browsers, and enterprise won't deploy an unapproved Chromium fork from a company best known for a search sidebar. The pricing architecture is almost certainly 'bundled into Perplexity Pro,' which means the browser is a retention mechanic, not a revenue line — that's fine until you realize the cost of maintaining a browser fork is not trivial and the ROI has to be measured in churn reduction, not new ARR. The moat question is the real problem: Chromium is open, the AI agent layer is replicable, and the switching cost for a browser is extremely high to create but fragile once created. This survives if Perplexity gets acquired by a platform player who needs an AI browser story; as a standalone business decision, the unit economics don't pencil.”

“The thesis Compass 2 is betting on: enterprise knowledge is fundamentally multimodal — it lives in slide decks, scanned contracts, financial tables, and annotated diagrams — and the first retrieval system that treats those formats as first-class citizens rather than edge cases will own the enterprise search layer. That's a plausible and falsifiable bet, but the dependency is that 'understanding' a chart means something semantically useful at query time, not just 'we embedded the image.' The second-order effect that matters here isn't faster document search — it's that if this works, structured data that currently lives locked in PDFs becomes queryable without a data engineering team to extract it, which shifts power from BI teams who own structured pipelines toward anyone with a document library. Cohere is riding the trend of on-premises LLM deployment for regulated industries — that trend is real and accelerating, and they're on-time to it, not early. The future state where this is infrastructure is 'every regulated enterprise has a Compass instance the same way they have an Active Directory instance.' I'd believe that in five years if the chart comprehension claim is real.”

“The thesis here is falsifiable: the browser is the last surface layer a model provider can own before cloud platforms commoditize the query layer, and whoever owns ambient web interaction owns the monetization stack that replaces the search ad. The dependency that has to hold is that users adopt a second browser for AI tasks — a behavior that has actually happened before with Arc, Brave, and Opera, so it's not implausible. The second-order effect nobody is talking about: if Comet's agent can observe full browsing context across sessions, Perplexity builds a behavioral dataset that no API-layer competitor can replicate, which is the real moat. The trend is browser-as-OS-layer, and Perplexity is early — not on-time, early — which means the execution risk is high but the position is genuinely differentiated.”

“The job-to-be-done is specific: execute multi-step web tasks without juggling tabs, extensions, and copy-paste loops — and that is a real job that knowledge workers hire for daily. The onboarding question is the one I can't answer from waitlist access, but the make-or-break moment is whether a user can complete a real task in the first five minutes without reading docs, because agentic products that require prompt engineering upfront die in onboarding. The completeness problem is that this requires switching your entire browser, which is a massive ask — Perplexity would have shipped a stronger product by nailing the extension first and using that install base as the migration funnel into Comet rather than leading with the browser. The specific product opinion I'd give them credit for: making the agent persistent and context-aware across the session, not just per-page, is the right call and meaningfully different from extension-based competitors.”