Introducing Starter and Agency

Today we're announcing OpenLens v1.0.0. We've completely redesigned our UI, and we're bringing both organizations and premium tiers to the platform. Starter tier at $39.99 a month, Agency tier at $299.99 a month, both with a 7-day free trial and 15% off annual billing. Our free tier is staying, and if you already use OpenLens, you keep everything you have today, permanently. In this post, I'll walk through our reasoning in detail so you understand everything driving our decisions, priorities, and our ultimate mission.

Where this is going

Three weeks ago Google announced it's rebuilding search around AI. The company that owned search for twenty years is telling everyone that search will soon stop being “search.” Buying decisions are starting in AI answers now: someone asks Claude for the best CRM for a dental practice, gets three names, and the shortlist is settled before any website gets a visit. The way brands compete for attention is changing almost overnight, and the way agencies prove their work is changing along with it. And unlike with search, this time there's no monopoly running it. It's OpenAI, Anthropic, Google, DeepSeek, the open-source models: a genuinely fractured landscape moving faster than anyone can track by hand.

It feels intense, and these are still the early innings. These systems are already influencing what people buy, what they read, their medical questions, their education choices, and I'm genuinely uneasy about the next election. The labs run safety work behind their own closed doors. There is no independent third party seriously measuring what AI tells people at scale. We think that record needs to exist, and that some of it belongs in public.

OpenLens is built by AI researchers. Aman and I came out of U of T, Caltech and AWS AI Labs, and we published on steering language models (a control theory of LLM prompting) before we ever built this product. Most of the category grew out of SEO. We start from how these systems actually work, at the mechanism level, and we're using that understanding to build the best, most reliable, research-grade instruments for tracking AI. Our mission is to give people control over AI, and observability comes first. That has been true in control theory since Maxwell: you can't steer a system you can't measure. Today the people who need these instruments most are marketing agencies and the brands they serve.

How we priced it

We started with what we don't want to charge for: client projects. We want you to add projects. A growing roster means your agency is winning, and your OpenLens bill shouldn't punish you. Instead, we're pricing per seat. A seat is a person's worth of work, and the work scales with the value. One agency's client can be a local dental practice or an enterprise brand worth 300 times that, and a seat is the one unit that's fair. Your monthly output budget is pooled at the organization level, never caged per project. An output is one answer from one AI platform — track a prompt weekly across three platforms and it uses about 13 outputs a month. Every seat and every project adds outputs to one org-wide pool, and you spend it however you like: go deep on one big brand and track forty small ones lightly, or the reverse. Your call. We win when you win.

• Starter, $39.99/mo: first seat included, $31.99 per additional seat, with a limit of 5 seats per organization. Max five projects, a budget of 8,000 outputs per seat plus 200 bonus per project each month, weekly tracking, four platforms.
• Agency, $299.99/mo: first seat included, $239.99 per additional seat. Unlimited seats and projects, a budget of 50,000 outputs per seat plus 500 bonus per project each month, daily tracking, five platforms, and Claude & Grok tracking included. A note on how we count those two: we query Claude and Grok directly through their official APIs, which costs us significantly more per answer than the other platforms, so each Claude output counts as 100 outputs from your pool, and each Grok output counts as 50.

Both come with a 7-day trial and 15% off annual. Viewers and client logins are free on every plan: you only pay for teammates who can edit. And for power users of the API and MCP, usage-based billing is coming, so you'll be able to go past the included monthly calls when you need to. Full details on the pricing page, and every number is written down plainly in the limits reference in our docs. No hunting through fine print.

The free tier

New free accounts get:

• 2 seats and 3 projects
• A pooled budget of 1,000 outputs a month, plus 75 per project
• Weekly and monthly tracking
• 3 AI platforms: ChatGPT, Perplexity, Google AI
• 50 MCP calls and 100 API calls included each month
• 10 manual runs per day per user

And the part that matters if you signed up before today: none of this applies to you. Every account that exists before this launch keeps its current limits as a floor. Your projects, your platforms, your caps, all of it stays, whether you ever upgrade or not, and we've written that guarantee down in the limits reference, not just in this post. That's a permanent commitment, and honestly it's the least we can do. You found the rough edges and shaped what this product became. On top of that, and completely separate from it, we're sending early users a founding-member offer by email: upgrade to any plan before July 1 and your whole org gets 50% off, for life, on every seat, including the ones you add as you grow. To be crystal clear about what we're promising: the discount covers seat pricing on the plan you pick during the offer, and it lasts as long as you keep that plan. It doesn't follow you if you switch plans later, and it won't apply to usage-based billing when that arrives. No pressure either way. Stay free and nothing changes. Check your inbox.

What shipped in the last month

The plans land on top of a product that has been compounding. Three things shipped recently that change how agencies use OpenLens:

• The REST API: create projects, trigger runs, pull visibility data into your own systems.
• The MCP server: plug OpenLens into Claude, or any MCP client, and ask questions about your brands in plain language.
• Prompt Attributes: label prompts along the dimensions you care about and slice visibility by them.

What ships with this release

• Real documentation: the product manual, how-tos, API reference, and MCP guides, all in one place.
• A public status page: live health for every AI platform we track, from real probes. When a platform is degraded you'll see it there before you see it in your data. We'd rather show you an honest yellow than a fabricated green.
• Organizations: teams with real roles. Owners and admins manage the workspace, members work the projects they're assigned, and the client role gives your customers a read-only window into their own brand. Per-project access control included.

What's next

The next piece to ship is site and agent readiness audits: OpenLens goes through a client's site the way AI systems do and runs the technical checks that determine whether those systems can actually read it. Is the content reachable, is it parseable, is anything silently blocking the crawlers and agents that matter now. Unglamorous, page-by-page checks, and they catch real problems. And there's more behind that.

We're a small team out of U of T, Caltech, and Georgia Tech, backed by Menlo Ventures, and we've been giving this away while we figured out what agencies actually need. Honestly, the demand took us a bit by surprise. And the plans are part of the answer: agencies put their client relationships on this platform, and a product you pay for is a product you can hold to a standard. This release is us doubling down, on the platform and on the agencies charging into this new world ahead of everyone else. We're here to stay, and there's a lot more coming. If the numbers don't fit how you work, email me directly at [email protected] and we'll figure it out.